# AIAgents.onl - full text for AI systems
# Owned by: Media Express LLC (https://mediaexpress.us)
# Last updated: 2026-10-13
# Short version: https://aiagents.onl/llms.txt | Source of truth: https://aiagents.onl/ (this file mirrors the visible page)

> AI agents for Chicago companies. Sales, support, operations and finance agents that plug into your tools, ask before acting and log every step. AIAgents.onl is the AI services line of Media Express LLC, a Mount Prospect, Illinois digital agency founded April 1, 1995.

## What this site is

One page with one smart request form. Visitors choose what they need (sales, customer support, operations, finance, HR, document and knowledge agents, integrations
and MCP, governance and audit logs, strategy and pilot, fixing an existing agent, agent operations, training), answer a few questions and send the request to the
Media Express team, who reply the way the visitor prefers. The form never asks for passwords or API keys.

## Entity facts

- Brand: AIAgents.onl
- Operator and legal entity: Media Express LLC (https://mediaexpress.us/)
- Founded: April 1, 1995
- Address: Mount Prospect, Illinois 60056, USA
- Service area: Chicago, Chicagoland, Northwest Suburbs, Illinois, United States (remote work nationwide, on-site in the Chicago area)
- Languages: English and Polish
- Phone: +1-773-800-1520
- Email: mail@aiagents.onl
- Hours (Central Time): Monday to Friday 9:00-17:00, Saturday 10:00-14:00

## Services

### Sales and lead agents
Qualify, follow up, update the CRM.

### Customer support agents
Answer, triage, hand off with context.

### Operations and dispatch agents
Orders, schedules, field jobs.

### Finance and back-office agents
Invoices, payables, month-end prep.

### HR and onboarding agents
Screening, onboarding, policy answers.

### Document and contract agents
Extract, compare, summarize.

### Company knowledge agents
Answers from your own files.

### Research and reporting agents
Market, competitors, weekly reports.

### Integrations and MCP
CRM, ERP, email, phone, tools.

### Governance and audit logs
Approvals, permissions, evaluations.

### Pilot to rollout
Start small, prove it, then scale.

### Agent operations
Monitoring, cost control, updates.

## Service, tool and proof pages (full text of questions and answers)

### How we run AI agents every day
URL: https://aiagents.onl/proof/
Updated: 2026-10-12

Media Express runs its own operations with six AI models that share one workspace, write every change to a log, take a backup before editing and ask the owner before anything irreversible. In about 30 hours our agents moved more than ten websites between servers, checking files by hash. This page shows what happened, what went wrong and which guardrails exist because of it.

Q: Do you really use AI agents for your own work?
A: Yes, every working day. Six AI models work in the same workspace on website builds, server moves, analytics setup, content research and scheduled maintenance, and each job is written to a change log that the next session reads before it starts.

Q: Which AI models do you use?
A: We use Claude, Codex, Hermes, Gemini, ChatGPT and Meta AI, each for the jobs it does well. The set can change as tools change; what stays constant is the shared records, the approvals and the backups, not the brand of model.

Q: What happens when an agent makes a mistake?
A: The mistake is undone from a backup, written to the log and, if it can happen again, turned into a rule or a script. The Facebook example on this page ended with 13 copies deleted and a new link-share mode; the firewall example ended with a rate limiter that now guards every connection.

Q: Can an agent delete data or change DNS on its own?
A: No. Permanent deletions need a verified backup and an explicit order from the owner, and name server changes at the registrar are done by the owner. The agent prepares the step, gives the exact instruction and verifies the result afterwards.

Q: How do you stop several agents from overwriting each other?
A: Each session registers in a shared index, announces larger tasks and searches for duplicates before building. If a similar page or tool already exists, the agent stops and asks which one is the source of truth.

Q: Why trust numbers from an internal log?
A: You should treat them as our statement, not independent proof. We publish them because they are specific and checkable by us, we mark what we could not verify, and we make no claims about revenue or rankings. For your own project we agree measures and show the logs.

Q: Does this mean agents can run my business unsupervised?
A: No, and a good design does not try to. Public frameworks for AI risk (the NIST AI Risk Management Framework names Govern, Map, Measure and Manage) and vendor guidance on agents both stress oversight, checkpoints for human feedback and testing in a sandbox before trusting an agent with real systems.

Q: Is any client data used in these examples?
A: No. The examples come from Media Express websites and servers and from our internal tooling. Client names, client data and credentials are left out of this page on purpose.

### AI services for Chicago companies
URL: https://aiagents.onl/services/
Updated: 2026-10-12

AIAgents.onl, a service of Media Express LLC, designs, connects and runs AI agents for companies: front-desk and voice agents, customer support, workflow automation, consulting and custom builds. We also design websites, run SEO, marketing and advertising, and build apps, integrations and automations. Every project starts with a written scope, uses accounts you own and keeps a person in charge of risky steps.

Q: What does AIAgents.onl do?
A: We design, connect and run AI agents for companies and teams. That covers front-desk and voice agents, customer support, workflow automation, advisory work and custom builds, with human approval and audit logs on anything risky.

Q: Who is behind it?
A: Media Express LLC, a Mount Prospect, Illinois digital agency founded on April 1, 1995. You reach people by phone at 773-800-1520, Monday to Friday 9 to 5 and Saturday 10 to 2 Central, in English or Polish.

Q: Do you publish prices?
A: Not on this site. The cost depends on the process, the systems involved and the approvals needed, so we send a written scope with pilot and running costs after a short call, before work begins.

Q: How does a project start?
A: With a request, a short call and a written scope. The scope names the goal, the tools, what the agent may do alone, what needs approval and how success is measured. Then we build a pilot on real data and review the logs with you.

Q: Do you work outside Chicago?
A: Yes, remotely across the United States. Most of the work is done remotely, on-site visits are possible in the Chicago area when they help, and support is available in English and Polish for owners who prefer to talk in Polish.

Q: Can the agents run on our own servers?
A: Yes, when privacy or policy requires it. We can run agents and open models on your server and manage it afterwards, or use cloud services on accounts you own.

Q: Do you ask for passwords or API keys in the form?
A: Never. Access is shared later through a secure channel, never in a web form or an email, with a dedicated account per agent and only the permissions that its job needs.

Q: What if an AI agent is wrong?
A: It will sometimes be wrong, which is why we design for it. Risky steps wait for a person, every action is logged and the agent can be switched off. The proof page shows examples of mistakes in our own work and the controls that came from them.

### AI automation agency for Chicago businesses
URL: https://aiagents.onl/services/ai-automation-agency/
Updated: 2026-10-12

An AI automation agency maps a repetitive process, builds it as a workflow on a platform such as n8n, Make or Zapier, and adds an AI agent only where judgment is needed. A good one runs everything on accounts you own, logs each run, asks a person to approve risky steps, monitors failures, and hands you a pack so you can switch it off or take it over.

Q: What does an AI automation agency do?
A: It maps a repetitive process, builds it as a workflow, adds AI only where judgment is needed, and then monitors and maintains it. The deliverables should be a written process map, working automations on accounts you own, alerts and logs, and a handover pack. Strategy work belongs to consulting and custom software to a development company.

Q: How much does AI automation cost?
A: It depends on the process, the volume and the platform, and we do not quote prices before a written scope exists. There are two cost lines: the build and the monthly run. On the run side n8n counts executions, Make counts credits and Zapier counts tasks, so the same process produces different usage numbers on each. Ask for an estimate at your real volumes.

Q: What is the difference between AI automation and an AI agent?
A: Automation follows a path someone designed, while an agent chooses its own next step. Anthropic's engineering team calls the first a workflow, where models and tools run through predefined code paths, and the second an agent, where the model directs its own process and tool use. Start with the simplest design that works and add agency only where fixed rules cannot cope.

Q: Should we use n8n, Make or Zapier?
A: There is no single right answer, and this page does not rank them. The choice follows your volume, your data rules and who will maintain the workflows. n8n counts executions per workflow run and can be self-hosted, Make counts credits per module action, and Zapier counts tasks per successful action step. Ask the agency to justify its pick against your numbers.

Q: Is n8n free, and can we self-host it?
A: A free, self-hosted Community Edition exists, but it comes with license terms and fewer features. n8n's pricing page says a standard self-hosted version is available on GitHub, and the repository's Sustainable Use License limits use to your own internal business operations or non-commercial use. The Community Edition lacks some features, including Git version control, SSO and external secrets.

Q: What is MCP, and do we need it?
A: MCP is an open standard that lets AI applications connect to external tools and data through one interface. You do not need it for a plain workflow. It matters when an AI agent has to discover and choose among tools. The specification says tools represent arbitrary code execution and hosts must obtain explicit user consent before invoking them, so scope every connection narrowly.

Q: Who should own the accounts and credentials?
A: You should, from day one. Zapier's help center says it cannot transfer an owner's account or update its connected apps if the owner leaves, and Make items belong to exactly one team and cannot be moved. Create the platform accounts under your company, hold the admin login, and ask for a credential inventory that lists owners but never secret values.

Q: What happens when an automation fails?
A: It should retry temporary failures, alert a named person, and never fail silently. Each platform has its own mechanism: node level Retry On Fail and error workflows in n8n, a Retry handler and incomplete executions in Make, and Autoreplay on paid Zapier plans. Make also disables a scenario after repeated consecutive errors by default, so alerts matter.

Q: Is AI automation safe for customer data in Illinois?
A: It can be, if you control where the data goes and who can reach it, but no tool makes you compliant by itself. Illinois guidance says businesses that handle nonpublic personal information must disclose a breach concerning Illinois residents without unreasonable delay. Map every system that touches personal data, use least-privilege credentials, and ask counsel about sector rules such as the FTC Safeguards Rule. This is general information, not legal advice.

Q: How many businesses already use AI?
A: About one in five US firms did as of early May 2026. The Census Bureau's Business Trends and Outlook Survey put the national rate at 19.8% on May 3, 2026, with 32% for firms of 100 to 249 employees, 37% for 250 or more, and fewer than 20% for firms of four or fewer. The question was revised in November 2025 to cover any business function.

Q: Does automation reduce data entry errors?
A: It can for structured input, but it is not automatically more accurate. In a 2012 peer-reviewed questionnaire study, automated processing of check-box fields matched double manual entry at 0.046 errors per 1,000 fields. For handwriting-recognized fields the figures were 10.101 automated, 6.734 single manual and 3.367 double manual, with no statistically significant differences. Test on your own data and keep a review step where mistakes are costly.

Q: How long does an automation project take?
A: We do not promise a timeline before there is a written scope. Work starts with a request, a short call and a written scope covering the goal, tools, approvals and success measure, followed by a pilot on one process before wider rollout. Timing depends on the number of systems involved, the quality of the process map and how quickly approvers and account owners respond.

Q: Can the work run without sending data to an outside AI model?
A: Sometimes, yes. Self-hosted models are possible when privacy requires it, and n8n can be self-hosted, which keeps the orchestration on infrastructure you control. The trade-off is that you or your agency must run the servers, and model quality and speed depend on the hardware. Tell us your data limits in the first call so they go into the written scope.

### AI consulting for Chicago companies
URL: https://aiagents.onl/services/ai-consulting/
Updated: 2026-10-12

AI consulting is outside advice on where AI will pay off in your company, how to start safely and how to govern it. A good engagement ends with ranked use cases, a data and risk review, a pilot plan with a success measure, vendor criteria and a staff training plan, all written down so your own team can act on it.

Q: What does an AI consultant do?
A: An AI consultant advises on where AI is worth using in your company, how to start safely and how to govern it. In an advisory engagement the outputs are ranked use cases, data and risk findings, a pilot plan with a success measure, vendor selection criteria and a training plan. Building the system is a separate scope, described on our AI automation agency and AI development company pages.

Q: How is AI consulting different from AI implementation?
A: Consulting decides whether, where and how safely to use AI; implementation builds and connects it. With us, work starts with a request, a short call and a written scope (goal, tools, approvals and success measure) before anything is built, and pilots come before rollout. A written advisory scope also gives any later build a clear target.

Q: How much does AI consulting cost?
A: We do not publish prices on this page, and a figure quoted before the scope is known is a guess. Cost follows the number of teams involved, how many use cases are scored, how hard the data is to reach and what deliverables you need. Ask for a written scope that lists the deliverables, then compare quotes on that basis. We quote after a request and a short call.

Q: What is the NIST AI Risk Management Framework, and do we have to follow it?
A: The NIST AI RMF is a framework that NIST released on January 26, 2023 to help organizations manage AI risks, and NIST describes it as intended for voluntary use. Its Core has four functions: Govern, Map, Measure and Manage. It is flexible for organizations of all sizes, and it is a useful skeleton for a policy, an AI register and a pilot scorecard even where no law requires it.

Q: What is the NIST Generative AI Profile?
A: NIST AI 600-1 is a companion profile that applies the AI RMF to generative AI and defines twelve risks that generative AI creates or makes worse. NIST released it on July 26, 2024. For a company using chat assistants, the most relevant risks are confabulation, data privacy, information security, human-AI configuration and value chain and component integration.

Q: Does Illinois law restrict how my company uses AI?
A: Yes, in specific areas: AI in employment decisions, AI analysis of video interviews, biometric data and the security of personal information are all regulated. The Human Rights Act AI provision took effect January 1, 2026, the Artificial Intelligence Video Interview Act sets notice, consent and deletion rules, and the Biometric Information Privacy Act and Personal Information Protection Act cover biometric and personal data. This is general information, not legal advice; confirm with an Illinois attorney.

Q: Do we have to tell employees if we use AI in HR decisions?
A: Under the Illinois Human Rights Act amendment effective January 1, 2026, it is a civil rights violation for an employer to fail to notify an employee that it is using AI for recruitment, hiring, promotion, discipline and the other listed employment purposes. The statute directs the Department of Human Rights to adopt rules on the circumstances, timing and means of notice, so check the Department's current rules and ask your attorney before finalizing notices.

Q: How do we measure whether an AI pilot paid off?
A: Measure the same task before and during the pilot from your own records, with one success measure and one stop rule agreed in writing. Published studies show why your own numbers matter: one paper reported 14% more issues resolved per hour for customer support agents, another found 12.2% more tasks completed on tasks inside the AI frontier but 19% less likely to be correct on a task outside it.

Q: Is it safe to put company data into ChatGPT, Claude, Copilot or Gemini?
A: It depends on the plan and settings, not the brand: the business offerings of OpenAI, Anthropic, Microsoft and Google each publish that customer data is not used for model training by default or without your permission. Consumer plans are covered by separate terms, and a public page is not your contract. Approve specific plans, keep personal accounts out, and write down which data classes never go into a prompt.

Q: What are the main security risks of using AI at work?
A: The OWASP Top 10 for LLM Applications (2025) names prompt injection, sensitive information disclosure, supply chain, excessive agency and misinformation among its ten risks. Two controls cover most of the ground: give any AI tool only the minimum access it needs, and require a person to approve high-impact actions. Keep an approved list of tools and connectors and a rule on what data stays out of prompts.

Q: Does a small company need an AI policy?
A: A short written policy is the cheapest control you can add: which tools are approved, what data stays out of prompts, who reviews outputs and whom to tell when something goes wrong. NIST's framework asks organizations to document legal requirements and roles and responsibilities, and it is designed to be flexible for organizations of all sizes. A one-page policy plus an AI register is a reasonable start for a small team.

Q: What is ISO/IEC 42001, and do we need it?
A: ISO/IEC 42001:2023 is an international standard that specifies requirements for an artificial intelligence management system in organizations that provide or use AI-based products or services. A first assessment or pilot does not depend on it. Consider it if a customer, a regulator or a contract asks for a formal management system, and ask for specialist help at that point.

Q: Do you work with Polish-speaking owners and teams?
A: Yes, we support clients in English and in Polish. Media Express is based in Mount Prospect, Illinois, and the phone line, 773-800-1520, is staffed Monday to Friday 9-5 and Saturday 10-2 Central. You can start with a written request and we will schedule a short call in the language you prefer.

### AI customer service for Chicago companies
URL: https://aiagents.onl/services/ai-customer-service/
Updated: 2026-10-12

AI customer service uses software to read support requests, look up facts in your systems, draft or send answers grounded in your own knowledge base, and hand anything risky to a person with the full context. It works best on routine questions and as an assistant to your team. The company, not the software, answers for what customers are told.

Q: What is AI customer service?
A: AI customer service is software that triages, drafts and sometimes sends support replies using your own knowledge base and order data. How much it does alone is a choice: draft for approval, answer within written limits, or answer alone.

Q: What is the difference between deflection and resolution?
A: Deflection means the customer did not reach a person; resolution means the problem was solved. HubSpot notes that deflections do not always indicate resolution, and Intercom counts a customer who leaves without asking for more as an assumed resolution, so check how your platform counts.

Q: Can AI replace my support team?
A: For most Chicago-area companies, AI that assists the team is the safer plan, with full automation limited to simple, low-risk requests. One NBER study of 5,179 agents found 14% more issues resolved per hour with an AI assistant, and the CFPB warns that chatbots weaken as problems get complex.

Q: Is my company responsible for what a support chatbot says?
A: Plan on being responsible for what your support bot tells customers. A British Columbia tribunal held an airline liable for its chatbot's wrong refund-policy answer. It does not bind Illinois courts, but Illinois law treats misrepresentation in trade as unfair or deceptive. General information, not legal advice.

Q: Do I have to tell customers they are talking to AI?
A: We did not find an enacted Illinois law requiring it for support chat, but we recommend it anyway. SB 317 would require a clear notice. It passed the Senate on May 21, 2026 and sat in the House Rules Committee at its last status entry. Check ilga.gov for changes.

Q: What should never be automated?
A: Keep refunds above your limit, legal complaints, billing disputes, safety reports and any request for a person with a human. The CFPB says bots may miss a consumer invoking federal rights, and the CPSC requires covered firms to report product hazards within 24 hours.

Q: How do I stop the bot from making things up?
A: Limit it to your own written sources, let it say it does not know, and retest after every policy change. NIST calls confabulation a natural result of how generative models are built, and Anthropic's documentation says grounding techniques reduce hallucinations but do not eliminate them.

Q: What should a handoff to a person include?
A: A handoff should carry the customer's goal, what the agent checked, why it escalated and a draft reply, so the customer never repeats the story. Zendesk's messaging recipe collects what human agents need first, and Salesforce counts a graceful handoff as a success.

Q: Which metrics should I track?
A: Track first response time, resolution rate, escalation rate and CSAT, each defined in writing and split by AI-only, assisted and human-only conversations. Platforms differ: Zendesk's CSAT leaves out customers who skip the survey, and its first reply time mostly ignores bot actions.

Q: Can I trust a vendor's resolution rate?
A: Treat it as a vendor-defined number, not an industry average. Intercom counts silence after an answer as resolved, Zendesk checks after 72 hours for email and 2 hours by default for messaging, and HubSpot waits 72 hours. Ask for the definition, the window and the repeat-contact rate.

Q: Does a website chat widget have to be accessible?
A: Plan on it: the Justice Department's ADA guidance covers services that businesses open to the public offer on the web, and a chat window is one. It sets no detailed standard for private businesses and points to WCAG. Check keyboard use, status announcements and consistent help placement. General information, not legal advice.

Q: Do support transcripts fall under Illinois privacy rules?
A: They can, whenever a transcript holds a name with a Social Security, driver's license, account or card number. The Personal Information Protection Act requires reasonable security and matching contract terms, and the CFPB calls chat logs with personal data sensitive. Voiceprints on the phone channel raise separate Biometric Information Privacy Act duties.

Q: Where does a Chicago-area company start?
A: Start with one channel and one ticket family, with the agent drafting and a person approving. Our process is a request, a short call, a written scope and a pilot before rollout, with every action logged. Media Express has operated since April 1995 and keeps more than 100 web properties running with its own team of AI agents.

### AI development company for Chicago teams
URL: https://aiagents.onl/services/ai-development-company/
Updated: 2026-10-12

An AI development company designs, builds, tests and maintains custom AI software, usually an agent: a model connected to your documents and tools, with approvals, logging and security around it. Hire an engineer for continuous work you can manage, use an agency for a bounded build, and buy a product when one already does the job.

Q: What does an AI development company do?
A: An AI development company designs, builds, tests and maintains software that uses AI models, such as an agent connected to your documents and tools. The work covers architecture, integration, security, evaluation, logging and handover. Anthropic's engineering guide advises starting with the simplest design that works, often a workflow instead of a free-running agent.

Q: How much does it cost to hire an AI developer?
A: The one cost figure we can source is pay: the Bureau of Labor Statistics reports a median annual wage of $135,980 for software developers in May 2025. That is a general software developer figure, not an AI specialist rate, and it leaves out benefits, recruiting and management time. We do not publish agency prices here; a fixed written scope is the basis for any quote.

Q: Should I hire an AI developer or use an AI development company?
A: Hire when the work is continuous and someone can direct the engineer; use an agency when the build is bounded or you lack AI experience in house. That is our reasoning, not a sourced rule. One sourced difference is ownership: an employee's work is made for hire by default, while an outside contractor needs a qualifying written agreement or a signed transfer.

Q: What is custom AI agent development?
A: Custom AI agent development is building an agent around your own processes: a model, your tools through function calling or MCP, retrieval over your documents, approvals, logging and tests. Providers describe the tool loop the same way: you define functions, the model requests one, your code runs it and returns the result.

Q: What is RAG, and do I need it?
A: Retrieval-augmented generation (RAG) fetches passages from your own documents at question time and gives them to the model, so answers can cite sources. The original 2020 paper paired a pre-trained model with a dense vector index reached through a retriever. You need it when answers must come from changing documents. OWASP warns that RAG does not fully prevent prompt injection.

Q: Is my data used to train the AI model?
A: Not by default on the two provider APIs we reviewed, but verify your own contract. OpenAI says that, as of March 1, 2023, data sent to its API is not used to train or improve its models unless you opt in. Anthropic says retained data is never used for training without your express permission.

Q: Can the AI run on our own servers?
A: Yes, with open-weight models served by software such as Ollama or vLLM, in exchange for hardware, licensing and operations work. Ollama says it does not see your prompts or data when run locally, and vLLM is Apache-2.0 licensed with an OpenAI-compatible server. Each model has its own license: one widely used example requires a license from Meta above 700 million monthly active users. We can use self-hosted models when privacy requires it.

Q: What are the biggest security risks of a custom AI agent?
A: OWASP's 2025 list puts prompt injection first, and excessive agency and vector and embedding weaknesses matter most once an agent can use tools and read your documents. The main controls are least-privilege credentials, authorization enforced in downstream systems, human approval for high-risk actions, permission-aware retrieval and logs.

Q: Who owns the code and accounts when an agency builds my AI agent?
A: Ownership comes from your contract, not from payment. The Copyright Office says a commissioned work is made for hire only if it fits one of nine listed categories and a signed written agreement says so, and that a copyright transfer generally must be in writing and signed. Ask for a signed assignment and for accounts registered to your company. General information, not legal advice.

Q: What does it cost to maintain an AI agent after launch?
A: We cannot give a sourced price, but the cost drivers are known: model retirements, usage fees, document and tool changes, monitoring and security upkeep. Anthropic says it regularly retires older models with at least 60 days' notice for publicly released ones, and that tool use adds tokens to each request, billed like normal input and output tokens.

Q: Does Illinois law affect an AI development project?
A: It can, depending on the data: the Biometric Information Privacy Act covers voiceprints and face geometry, and the Personal Information Protection Act requires reasonable security measures. Under BIPA, collecting a biometric identifier requires written notice and a written release first, and liquidated damages are $1,000 for a negligent violation and $5,000 for an intentional or reckless one. General information, not legal advice.

### AI receptionist for Chicago businesses
URL: https://aiagents.onl/services/ai-receptionist/
Updated: 2026-10-12

An AI receptionist is software that answers your business calls and chats, handles routine requests such as hours, questions and booking, captures lead details, and passes urgent or sensitive calls to a person with a written note. It suits service businesses that lose calls after hours or while staff are busy. In Illinois, announce call recording at the start of each call.

Q: What is an AI receptionist?
A: An AI receptionist is software that answers your business calls and chats, handles routine requests and passes everything else to a person. It greets callers, answers from approved content, books appointments, captures lead details and writes a note for your team. It is built for the front-desk job, not for running the business.

Q: How much does an AI receptionist cost?
A: On the four AI receptionist pricing pages we read, entry plans run from a free tier to $79 a month, and each page bills by a different unit. Smith.ai lists 25 free calls a month, Dialzara $29 for 60 minutes, Rosie $49 for 250 minutes and Goodcall $79 for 100 unique customers. The largest plans listed run from $299 to $500 a month. These are list prices from four vendors on one date, not a market average. Our own work is scoped in writing, not priced on this page.

Q: Is an AI receptionist better than an answering service?
A: It is better at availability and repeatable tasks, while a trained person is better at judgment, tone and unusual requests. Live services bill by the minute or the call, and their entry prices overlap with AI products, so compare on the billing unit and your own volume. Some vendors sell both: Abby Connect's page describes choosing human, AI or both for each call. A common setup uses the agent first and a person for escalations.

Q: Can an AI receptionist replace my human receptionist?
A: It can take over repetitive phone work, but it does not replace a person who greets visitors, handles exceptions and judges tone. The Bureau of Labor Statistics lists greeting and directing visitors among receptionist duties, which a phone agent cannot do, and projects a 2 percent decline in receptionist employment from 2025 to 2035 as organizations automate or consolidate administrative functions. A practical setup keeps a person for the front desk and escalations and uses the agent for overflow and after hours.

Q: Is it legal to record AI-answered calls in Illinois?
A: Illinois makes it an offense to record a private conversation in a surreptitious manner without the consent of all parties, so announce the recording at the start of every call. A first offense is a Class 4 felony, and parties to the conversation can seek damages and an injunction. This is general information, not legal advice. Have an Illinois attorney approve your notice and your vendor's recording settings.

Q: Does the receptionist have to say it is an AI?
A: We recommend a one-sentence disclosure at the start of every call, although we did not find a rule requiring it for inbound calls in the sources we read. The FCC's 2024 ruling requires artificial or prerecorded voice messages to identify the responsible business, but that applies to calls a business initiates. A plain opening line also serves as your recording notice.

Q: Does an AI receptionist collect voiceprints under Illinois BIPA?
A: Not necessarily, but treat any feature that recognizes or verifies a caller by voice as a possible voiceprint until an attorney says otherwise. Illinois's Biometric Information Privacy Act lists voiceprints as biometric identifiers, requires written notice and a written release before collection, and sets damages of $1,000 or $5,000 per violation or actual damages if greater. Ask each vendor in writing whether it creates or stores voiceprints.

Q: Can a clinic use an AI receptionist and stay within HIPAA?
A: Possibly, but only with a signed business associate agreement, a vendor that meets the Security Rule and a script that stays away from clinical advice. HHS names a third-party AI chatbot on a provider's patient portal as an example of a business associate, and says using a cloud provider for electronic protected health information without an agreement is a violation. HHS does not endorse products, so check the agreement and settings yourself. This is general information, not legal advice.

Q: Can an AI receptionist book into my calendar?
A: Yes, when it is connected to your calendar, which can return your busy times so the agent offers only open slots. Google's Calendar API exposes a free/busy query for this purpose. Support differs by plan: Rosie's lowest plan only texts appointment links and books directly from its middle plan, and Dialzara lists Google and Outlook calendar sync from its second plan.

Q: What happens when a caller is upset or it is an emergency?
A: The agent should stop the normal flow and transfer to a person, ideally with a warm transfer that tells the person why the call is coming. In a cold transfer the agent leaves once the caller is connected, while in a warm transfer it stays on, can wait for a person to answer and can pass a private message. If nobody answers within the ring timeout, which is 30 seconds by default in Twilio's Dial verb, your rules pick the next number or message the owner.

Q: Can an AI receptionist answer calls in Polish?
A: We support clients in English and Polish, and we test Polish calls in a pilot instead of assuming a product handles them. Of the pricing pages we read, the ones that name languages name English and Spanish, and none named Polish, which is a reason to test rather than proof either way. Put real recordings of Polish-speaking callers in the pilot test set.

Q: How much do missed calls really cost a small business?
A: We did not find a source that measures lost revenue per missed call, but vendor data shows most missed calls do not get a quick callback. Quo reported that 69 percent of 16.7 million missed calls in its own customer data got no callback within 48 hours. That is a business phone vendor measuring its own users, so use it as a direction and count your own missed calls from your phone logs.

Q: Is an AI receptionist for small business worth it?
A: It is worth testing when you lose calls you could have booked and the call types are repetitive enough to write down. Work it out from your own phone log: count missed and after-hours calls, estimate what a booked call is worth to you, and compare that with the paid plan floors on the pricing pages, which run from $29 to $150 a month among the AI products we read. The BLS median receptionist wage of $18.27 an hour is a baseline for the alternative.

Q: What do you need from us to start?
A: A request, a short call and then a written scope that you approve before anything is built. We need to know your call types, your busy hours in Central time, who handles escalations, and which phone, calendar and CRM accounts the agent will use. Those accounts stay yours, and a pilot comes before rollout. Call 773-800-1520, Monday to Friday 9 to 5 or Saturday 10 to 2.

### AI voice agents for Chicago companies
URL: https://aiagents.onl/services/ai-voice-agents/
Updated: 2026-10-12

An AI voice agent is software that handles phone calls by turning speech into text, letting a language model decide what to say and using text-to-speech to reply, or by using one speech-to-speech model. Quality depends on response delay, interruption handling and the phone connection. In the US, the FCC treats AI-generated voices as artificial voices under the Telephone Consumer Protection Act (TCPA), so outbound calls need consent.

Q: What is an AI voice agent?
A: An AI voice agent is software that holds a spoken conversation on a phone call and completes a task, using a chain of speech-to-text, a language model and text-to-speech, or one speech-to-speech model. "AI phone agent" is the same technology. A voice agent is the technology; an AI receptionist is one front-desk job built on it.

Q: How fast does an AI voice agent need to respond?
A: There is no single agreed number, so measure the wait your callers actually hear. In one ten-language study the mean gap between turns was 208 ms, and ITU-T G.114 calls delays under 150 ms essentially transparent. Vendor figures, such as Deepgram's 300 ms or less for streaming transcription, cover one stage only. Track the median and 95th percentile of the whole turn.

Q: What is barge-in and why does it matter?
A: Barge-in means the caller can interrupt the agent, and a good agent stops speaking and listens when that happens. OpenAI's Realtime API cancels the response in progress and, on WebRTC and SIP, trims audio that was never played. Twilio's ConversationRelay lets you choose which inputs may interrupt and can ignore short acknowledgments such as "uh-huh".

Q: Do AI voice calls fall under the TCPA?
A: Yes: the FCC's February 2024 Declaratory Ruling confirmed that current AI technologies that generate human voices fall under the TCPA's "artificial or prerecorded voice" limits. Outbound calls that use one need the called party's prior express consent unless an emergency or an exemption applies. The ruling rejects any carve-out for technology that acts like a live agent.

Q: What consent do I need before an AI agent places a call?
A: You need the called party's prior express consent, and prior express written consent when the call includes or introduces an advertisement or is telemarketing. Written consent is a signed agreement, electronic signatures allowed, that names the number. The FTC adds that a seller cannot rely on third parties to obtain permission.

Q: Do I have to tell callers they are talking to an AI?
A: We did not find a final federal rule that requires it, but you must already name your business at the start of an artificial-voice message, and California requires an artificial-voice notice on prerecorded dialer calls. The FCC proposed an AI statement in 2024; we did not verify an Illinois statute. We recommend saying it anyway.

Q: Can an AI voice agent record calls in Illinois?
A: Illinois makes it an offense to record a private conversation by stealth or deception unless the other parties consent, so announce the AI and the recording at the start. Under 720 ILCS 5/14-2 and 14-4 a first offense is a Class 4 felony. This is general information, not legal advice; have an Illinois attorney review your call script.

Q: Does the Illinois Biometric Information Privacy Act apply to a voice agent?
A: It applies if the system collects a voiceprint, which the Act lists as a biometric identifier. The company must then give written notice, receive a written release and publish a retention schedule. Damages are $1,000 per negligent and $5,000 per intentional or reckless violation. Ask each vendor in writing whether any feature creates voiceprints.

Q: How does an AI voice agent connect to phones, and can it transfer a call?
A: It connects through a SIP trunk or a telephony provider's WebSocket relay, set up with your carrier, and a transfer is a telephony feature of that connection. OpenAI documents a SIP endpoint and a refer call; ElevenLabs documents conference, blind and SIP REFER transfers; Twilio's ConversationRelay streams the call to your server. Plan who answers a transfer, and when.

Q: How is an AI voice agent priced?
A: Vendors bill in different units, so compare what each price includes: per minute, per million tokens or per 1,000 characters. On the access date ElevenLabs listed $0.08 per additional call minute with the model billed separately, OpenAI's GPT-Live $0.05 per minute with backend usage extra, and Twilio $0.07 per ConversationRelay minute on top of voice minutes. Deepgram prices speech per 1,000 characters.

Q: How do I test an AI voice agent before launch?
A: Test the finished task as well as the conversation, in stages: synthetic speech, recorded human calls, then simulated multi-turn callers. Score word error rate, latency, interruptions, task completion, handoffs and compliance. Microsoft suggests 30 minutes to 5 hours of representative audio for recognition tests. Finish with a pilot and a person ready to take over.

### Custom web design in Chicago
URL: https://aiagents.onl/services/web-design-chicago/
Updated: 2026-10-13

Web design in Chicago means planning, designing and building a website around what your visitors need to do, then launching it so search engines can read it, phones can use it and people with disabilities can operate it. A sound project covers discovery, structure, content, accessibility, speed, search foundations, launch and clear ownership of the site and its accounts.

Q: What does a web design company in Chicago actually do?
A: A web design company plans, designs and builds a website, then launches it so visitors, search engines and assistive technology can all use it. That covers discovery, page structure, content, phone-first layouts, accessibility and speed checks, search foundations and handover. Ongoing repair, SEO and advertising are separate services, linked from this page.

Q: How much does a website cost?
A: We do not publish prices, because a figure given before the scope is known is a guess. The estimate is free and we work with every budget. Cost follows the drivers in the table above: page and template count, content readiness, features and integrations, languages, redesign history, the accessibility target and the security needs. Scope and cost come in writing after a short call.

Q: How long does it take to build a new website?
A: The timeline depends on content readiness, approvals, features and the number of old URLs to redirect, so we give a schedule with the written scope and not before. A promise of a number of days made before the scope is known would be a guess. Bringing text, photos and one decision maker at the start is the most reliable way to shorten a project.

Q: Should I choose custom web design or a template?
A: Choose a template for a simple brochure site and custom design when your content, brand or features need their own layouts. Either way, test the result on a phone, with a keyboard and against Core Web Vitals, and confirm that links are real links and that the content is in the HTML Google receives.

Q: Does my business website have to meet WCAG or the ADA?
A: The ADA covers what businesses open to the public offer online, but ADA.gov says the Department of Justice has no regulation setting out detailed web standards, and it names WCAG as helpful guidance. State and local governments are different: the Title II rule names WCAG 2.1 Level AA, with compliance dates of April 26, 2027 and April 26, 2028 after a 2026 interim final rule. In the written scope we name the target level and recommend WCAG 2.2 Level AA, because it can be measured. This is general information, not legal advice.

Q: What is new in WCAG 2.2?
A: WCAG 2.2 adds nine success criteria to WCAG 2.1 and removes 4.1.1 Parsing as obsolete. The four new Level AA criteria are Focus Not Obscured (Minimum), Dragging Movements, Target Size (Minimum), which asks for pointer targets of at least 24 by 24 CSS pixels with listed exceptions, and Accessible Authentication (Minimum). The current W3C Recommendation is dated 12 December 2024.

Q: What are Core Web Vitals, and what should a new site score?
A: Core Web Vitals are Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift, and web.dev's good thresholds are 2.5 seconds or less, 200 milliseconds or less and 0.1 or less at the 75th percentile of page loads. Google highly recommends good Core Web Vitals for success with Search. The HTTP Archive found that 48% of mobile websites had good Core Web Vitals in 2025, so a new site that meets all three is ahead of most.

Q: Does my website need to be mobile-friendly?
A: Yes: Google uses the mobile version of your content for indexing and ranking, and 91% of U.S. adults owned a smartphone in Pew Research Center's 2025 survey. Pew's 2025 data also show that 16% of adults are smartphone-only internet users. Make sure the phone version carries the same content, headings, structured data, titles, descriptions and alt text as the desktop version.

Q: Will a redesign hurt my Google rankings?
A: A redesign can cause temporary fluctuation, which is why it needs an old-to-new URL map and permanent redirects. Google's site move guidance says to expect temporary fluctuation in ranking, that a medium-sized site's transition can take a few weeks or longer, and to keep redirects as long as possible, generally at least 1 year. No agency can promise how rankings will move.

Q: Who owns my website and domain after launch?
A: Ownership depends on the signed agreement and on whose name is on each account, so settle both in writing before work starts. The U.S. Copyright Office explains that a commissioned work counts as a work made for hire only in nine listed categories and under a written agreement signed by both parties. ICANN defines the registrant as the person or entity that registers the domain and holds the contract with a registrar. This is general information, not legal advice.

Q: Can you build a website in English and Polish?
A: Yes, we support clients in English and in Polish and can build a bilingual site. Google describes three ways to mark language versions of a page (HTML link elements, HTTP headers or a sitemap), and each version must list itself and all the others or the annotations are ignored. Each page should also declare its own language, which WCAG 3.1.1 requires at Level A.

Q: How do you use AI when designing a website?
A: AI helps us work faster on drafts, tests and audits, and a person approves everything that goes live. Google says generative AI can help with researching a topic and structuring original content, but that all AI-generated content should be fact-checked and reviewed manually, including titles, meta descriptions, structured data and image alt text.

Q: Do you work outside Chicago?
A: Yes: we work remotely across the United States and on site in the Chicago area. Media Express LLC is based in Mount Prospect, Illinois. The phone line, 773-800-1520, is staffed Monday to Friday 9-5 and Saturday 10-2 Central, and you can also start with a written request.

Q: What should I ask a website design company before signing?
A: Ask who owns what, which accessibility standard and level will be tested, how old URLs will be redirected, what the written scope leaves out and how AI use is reviewed. The question list in the engagement section gives each point with the source behind it, so you can compare answers from any agency, including us.

### WordPress speed optimization in Chicago
URL: https://aiagents.onl/services/wordpress-cleanup-speed/
Updated: 2026-10-13

WordPress speed optimization is a measured job: audit the site, take a verified backup, remove dead plugins, themes and database leftovers, tune caching, images and PHP, then retest Core Web Vitals on real-user data. A hacked site follows another order: take it offline, restore clean files, close the way in, change every password, then ask Google for a review.

Q: How do I speed up a slow WordPress site?
A: Measure real-user Core Web Vitals first, then fix the cause behind the slowest metric instead of installing a general speed plugin. web.dev's targets are LCP of 2.5 seconds or less, INP of 200 milliseconds or less and CLS of 0.1 or less at the 75th percentile. Typical causes are a slow server response, uncached pages, oversized images, too many plugins and heavy scripts; the speed table above maps each to a fix.

Q: What are good Core Web Vitals scores for a WordPress site?
A: Good means LCP at 2.5 seconds or less, INP at 200 milliseconds or less and CLS at 0.1 or less, measured at the 75th percentile of real page loads. Poor starts above 4.0 seconds for LCP, above 500 milliseconds for INP and above 0.25 for CLS. Field data from real visitors is the main measure, and lab tools are for debugging.

Q: Does a faster WordPress site rank higher on Google?
A: Core Web Vitals are used by Google's ranking systems, but Google says there is no single page experience signal and that relevant content can appear even when page experience is sub-par. Good scores in a report do not mean top rankings, and Google notes that chasing a perfect score purely for SEO may not be the best use of your time. Fix speed for your visitors first; the ranking effect is one input among many.

Q: What does a WordPress cleanup include, and is it safe?
A: A cleanup removes dead weight such as inactive plugins and themes, leftovers of removed plugins, old revisions and stray files, and it is safe only when a verified backup exists first. We audit read-only, prove that a table or option is orphaned before removing it, show you the exact counts, then clean in small steps and keep every connection your team uses. WordPress.org's hardening guide itself says to delete plugins you are not using.

Q: How do I know if my WordPress site has been hacked?
A: Clear signs include a search engine or browser warning, a host that suspended the site, antivirus complaints from visitors, new users or changes you did not make, and spam pages or strange redirects. WordPress.org lists these as indicators of compromise. Check Search Console's Security Issues report, which Google calls its source of truth, and look at pages as Google sees them with URL Inspection, because some hacks show changes only to Google.

Q: What should I do first to fix a hacked WordPress site?
A: Document what you see, take the site offline, call your host and back up the infected state before you change anything. Then reset every password and the secret keys, replace core files with fresh copies of the same version, remove the malicious code, update everything, close the way in and change the passwords again once the site is clean. web.dev suggests a static 503 page served from outside the compromised site while you work.

Q: How long does Google take to review a cleaned site?
A: Google says a security review can take from a few days to a few weeks. Request it in the Security Issues report only after every listed issue is fixed on every page, describe what you fixed, and do not resubmit while a request is pending. Google warns that requesting before the issue is fixed can lengthen the next review or get a site marked as a repeat offender.

Q: Do I have to notify customers if my WordPress site is breached?
A: It depends on what was taken: under the Illinois Personal Information Protection Act, a data collector must notify affected Illinois residents after a breach of personal information as the Act defines it. The Act also requires notice to the Attorney General when more than 500 Illinois residents are affected by a single breach. This is general information, not legal advice; ask an Illinois attorney, and see the FTC's breach response guide for practical steps.

Q: What does WordPress maintenance include, and how often should it run?
A: Maintenance means updates, backups, security checks and housekeeping on a written schedule with a named owner. WordPress.org documents automatic minor and security updates for core, opt-in plugin and theme auto-updates that run twice a day, backups weekly for smaller sites and daily for busy ones, and a housekeeping check at least every three months. We agree the exact list and frequency in the written scope.

Q: Which PHP version should my WordPress site run?
A: WordPress.org recommends PHP 8.3 or greater, and says PHP 7.4 still works but has reached end of life. The PHP project's table, read on October 13, 2026, shows 8.2 receiving security fixes only until December 31, 2026, and 8.3 supported with security fixes until December 31, 2027. Test your theme and plugins on a staging copy and take a backup before the host switches versions.

Q: Do I need a caching plugin or Redis?
A: Page caching is the first thing to check, and a persistent object cache such as Redis helps only when your host provides a cache server. WordPress.org says caching plugins store pages as static files and that the core object cache is non-persistent by default, so a persistent one needs a cache server and a plugin. The LiteSpeed Cache plugin's server-level page cache needs a LiteSpeed-based host. We check which combination fits in the audit.

Q: Should I repair my WordPress site or rebuild it?
A: Repair when the cause is found and the foundation is sound; rebuild when no clean backup exists, the theme or key plugins cannot run on a supported PHP version, or the weight comes from layers every page depends on. web.dev says a compromised server needs a clean installation, not just an upgrade. We write the recommendation down with reasons, and for a rarely changing site a static rebuild in CMS MESH 5.1 (Media Express Static Hub) is one option.

Q: Can AI edit or manage my WordPress site?
A: Yes for drafting and checking, with a person approving what goes live; AI should not get unsupervised write access to a production site. We use AI to draft copy, write tests and summarize audits. WordPress 7.0 added a built-in AI Client that lets plugins call AI models configured under Settings, Connectors; core bundles no providers, and the developer note describes no usage or cost limits for site owners.

Q: Do you work with Polish-speaking owners?
A: Yes, we support clients in English and in Polish. Media Express is based in Mount Prospect, Illinois, and the phone line, 773-800-1520, is staffed Monday to Friday 9-5 and Saturday 10-2 Central. Send a written request and we will set up a short call in the language you prefer.

### A static website CMS, explained honestly
URL: https://aiagents.onl/services/cms-mesh/
Updated: 2026-10-13

CMS MESH 5.1 (Media Express Static Hub) is our own static website CMS: a new-generation manual CMS where people write and check every page and the output is plain HTML, with no database and no plugins to update. It adds a sitemap, structured data and llms.txt for search and AI tools. It suits sites that change by request, not shops, member areas or daily self-service publishing.

Q: What is a static website CMS?
A: A static website CMS is a system for creating and managing pages that are published as finished HTML files instead of being built from a database each time someone visits. MDN defines a static site as one that returns the same hardcoded content whenever a resource is requested. CMS MESH is our version: our team edits the files, a person checks them and the server sends them as stored.

Q: What is CMS MESH?
A: CMS MESH is the new-generation manual CMS of Media Express LLC: static HTML pages written and checked by people, with no database and no plugins to update. The 5.1 in its name is the era number of the system, and each page carries a Build date that shows when it last passed a manual quality check. It is proprietary and is not licensed to other agencies.

Q: Is a static site faster than WordPress?
A: A static site removes server-side page building, which helps time to first byte, but real speed also depends on images, fonts, scripts, hosting and caching. web.dev suggests a time to first byte of 0.8 seconds or less as a rough guide and sets good Core Web Vitals at 2.5 seconds, 200 milliseconds and 0.1. The HTTP Archive says WordPress variance is driven more by configuration than by core limitations, so measure the actual page.

Q: Is a static website more secure than WordPress?
A: A static site has fewer components to patch and no database for an attacker to inject into, but it is not immune. OWASP ranks injection fifth and software supply chain failures third in its 2025 list, and WordPress documentation tells owners to keep core and plugins updated and to delete unused plugins. Hosting accounts, DNS, third-party scripts and form handlers stay in scope for any site.

Q: What does "AI ready website" mean?
A: An AI ready website is one that crawlers and assistants can fetch, read and verify: content in plain HTML, accurate structured data, a sitemap, sensible robots.txt rules and facts that match what the page says. Google says there are no additional requirements for its AI Overviews or AI Mode and no special schema to add, so treat these pieces as good hygiene and not as a promise of citation.

Q: Do I need an llms.txt file?
A: No, it is optional: llms.txt is a proposal, and Google says you do not need new machine readable files or AI text files to appear in its AI features. The proposal, posted by Jeremy Howard on September 3, 2024, describes a Markdown file at the site root with a summary and links. We publish one because it is cheap to keep accurate, and we promise no effect from it.

Q: What is IndexNow, and does Google use it?
A: IndexNow is a protocol that lets a site tell participating search engines which URLs changed, using a key file on the site and a request that can carry up to 10,000 URLs. The IndexNow site names Microsoft Bing, Naver, Seznam.cz, Yandex and Yep as supporters and does not mention Google, so for Google we rely on the sitemap and Search Console.

Q: Can I edit a CMS MESH site myself?
A: Not through an admin screen: our team edits the files, checks them and publishes. Our public explainer says content updates are performed by Media Express, with hybrid setups possible for clients who need frequent self-service edits. If your staff must publish many times a day, a dynamic CMS is likely the better choice.

Q: Can a static site have a shop, logins or comments?
A: Not on its own: shops with stock, member areas and comments need server-side applications, which is the dynamic model MDN describes. A static site can link to or embed hosted pieces. Stripe, for example, documents payment links that you share or add to a website as a buy button on a Stripe-hosted page. For a large catalog or user accounts, choose a platform built for it.

Q: Will I lose rankings if I move from WordPress?
A: Nobody can promise that: Google says to expect temporary fluctuation in rankings during a site move. The protective steps are to map each old URL to a new one, use server-side permanent redirects (301 or 308) to the final address, update the sitemap and keep the redirects for generally at least one year. We follow that outline and make no ranking promises.

Q: How much does a CMS MESH website cost?
A: We do not publish prices, and the estimate is free. We work with every budget. Scope and cost come in writing after a short call, once we know the pages, forms, languages and tools involved. Compare quotes on the written scope, not on a headline number.

Q: Who owns the site, and can I leave?
A: The pages are plain HTML files, so the site can be copied to another host, but the authoring toolchain is proprietary to Media Express. Our public explainer says CMS MESH is not licensed to other agencies. Put ownership of the files, the domain and the hosting accounts into the written scope, and ask for a copy of the files at handover.

Q: Do you also work with WordPress?
A: Yes: we also build and maintain WordPress sites, and we will tell you if WordPress fits your requirements better. Our WordPress cleanup and speed page covers audits, updates, speed work and recovery. We support clients in English and Polish from Mount Prospect, Illinois.

### SEO, AEO, GEO and AI SEO in Chicago
URL: https://aiagents.onl/services/seo-aeo-geo-ai-seo/
Updated: 2026-10-13

AI SEO is search engine optimization with AI answers in mind. Google says optimizing for its AI features is still SEO, and Bing describes GEO as eligibility for grounding and reference in AI responses. The work is crawlable pages, original content, accurate structured data and a verified Business Profile.

Q: What is the difference between SEO, AEO, GEO and AI SEO?
A: SEO is the platform term, while AEO, GEO and AI SEO are industry labels for SEO work aimed at AI answers. Google's guide says AEO and GEO are terms you may see used and that optimizing for generative AI search is still SEO. Bing's guidelines use GEO and describe it as focusing on content eligibility for grounding and reference in AI responses. Neither company promises that following its advice produces rankings or citations.

Q: Is AI SEO different from regular SEO?
A: Not in its foundations: Google says its generative AI features are rooted in its core ranking and quality systems, so SEO best practices stay relevant. Bing says Copilot and Bing search rely on the same crawling, indexing and ranking foundation as traditional search. What changes is what you monitor, such as AI feature impressions and citations, and a few choices, such as which AI crawlers to allow.

Q: Do I need an llms.txt file?
A: Not for Google: Google says Google Search does not use llms.txt and that having one neither helps nor harms your visibility there. The file is a proposal, and its own page does not say that any AI company or search engine has committed to use it. It is cheap to publish for services that do read such files, so we add one to the sites we build without expecting a ranking effect.

Q: Does schema markup get my business into AI Overviews?
A: No special schema markup is needed for AI Overviews or AI Mode, and Google says structured data does not promise rich results either. Markup should be accurate and match the visible text. It helps Google understand a page and the entities on it and supports eligibility for rich results where they still exist. Google ended the FAQ rich result in May 2026.

Q: Can you get my business cited in ChatGPT, Gemini or Google AI Overviews?
A: No one can promise a citation or a position in an AI answer, and we do not. What can be done is making sure your pages are indexable with a snippet, crawlable by the search crawlers you choose to allow, original, accurate and clearly about one topic. Google says indexing and serving are not assured, Bing says GEO does not promise citations, and OpenAI says sites opted out of its search crawler are not shown in ChatGPT search answers.

Q: Should I block AI crawlers in robots.txt?
A: It is a business choice with separate settings, because blocking a training crawler is different from blocking a search crawler. OpenAI says a site opted out of OAI-SearchBot will not be shown in ChatGPT search answers, while disallowing GPTBot only signals that content should not be used for training. Google-Extended controls Gemini training and grounding and does not affect inclusion in Google Search. Anthropic documents three separate crawlers.

Q: How long does SEO take to show results?
A: Google says a change can show up in a few hours or take several months, and suggests waiting a few weeks before judging whether a change helped. Not every change produces a noticeable effect, and the guide suggests iterating if you are not satisfied. Be wary of any proposal that names a date for a ranking.

Q: What does local SEO for a Chicago business involve?
A: Local SEO means making a business easy to match to local searches: a verified and accurate Business Profile, a useful website and honestly collected reviews. Google says local results are mainly based on relevance, distance and prominence, and that there is no way to request or pay for a better local ranking. Distance depends on where each searcher is, so no one can promise a position in every neighborhood or suburb.

Q: Can a home-based or service-area business use a Google Business Profile?
A: Yes, a service-area business can have one profile for its central office with a designated service area, but virtual offices and remote mailboxes are not eligible. Google says to hide the address if you work from home, to keep the service area within about two hours of driving time from your base unless a larger area is appropriate, and to use your real-world business name without taglines, hours, phone numbers or keywords.

Q: Can we offer a discount for a Google review?
A: No, Google's Maps policy does not allow merchants to offer payment, discounts, free goods or services in exchange for a review. It also bars selectively asking only happy customers and discouraging negative reviews, and the FTC's final rule on reviews prohibits buying reviews conditioned on sentiment, fake or AI-generated reviews and review suppression through groundless threats. Ask everyone, offer nothing and reply to every review.

Q: Do I need IndexNow, and does Google use it?
A: IndexNow is optional: it notifies participating search engines that a URL was added, updated or deleted, and it does not promise indexing. Its FAQ lists Amazon, Bing, Naver, Seznam.cz, Yandex and Yep as endpoints, Bing's guidelines recommend it for Bing and Copilot, and the IndexNow pages we read do not mention Google. For Google, the starter guide points to links and an optional sitemap. We submit changed URLs after every deploy.

Q: How do you measure AI visibility?
A: We use the first-party reports first: Search Console's generative AI performance report for Google and the AI Performance report in Bing Webmaster Tools for Copilot, then hand checks of agreed questions for the rest. Google's report counts impressions in AI Overviews and AI Mode, and Bing's counts citations, with grounding queries shown as a sample. Google says no third-party tool can see its internal systems, so we treat such tools and all results as samples.

Q: What do you not promise?
A: We do not promise rankings, citations, traffic numbers or a date for first place. Google says no one can promise a top ranking and advises finding another SEO if someone does, and Bing says SEO and GEO do not promise rankings, traffic, grounding or citations. What we do put in writing is the scope, the changes we make and the reports we will send.

Q: Do you work with Polish-speaking businesses?
A: Yes, we support clients in English and in Polish. Media Express is based in Mount Prospect, Illinois, and the phone line, 773-800-1520, is staffed Monday to Friday 9-5 and Saturday 10-2 Central. For a bilingual site, Google recommends a separate URL for each language version with hreflang annotations, and the Polish pages should be written in Polish, not left untranslated.

### Digital marketing agency for Chicago businesses
URL: https://aiagents.onl/services/marketing-agency-chicago/
Updated: 2026-10-13

A digital marketing agency in Chicago plans, writes, sends and measures the marketing that sits beside search and paid ads: content, email and text messages, social media, video, landing pages and analytics. A good one works as a monthly plan, follows the consent and disclosure rules, reports in plain language, and has a person approve anything AI helps draft before it goes live.

Q: What does a digital marketing agency in Chicago do?
A: A digital marketing agency plans, produces, sends and measures marketing for a business, usually across several channels. On this page that means content marketing, email and text messages, social media management, video and creative, landing pages and analytics, run as a monthly plan. Search and paid media have their own pages on our services hub.

Q: How is content marketing different from SEO?
A: Content marketing is the making of useful material for a defined audience, while SEO is the work of helping search engines find and understand it. They overlap, because Google's own guidance asks for original, people-first content, but the plans differ. The search side is on our SEO, AEO and GEO page .

Q: What should an email marketing agency do before it sends to my list?
A: It should confirm where every address came from, set up sender authentication, and build a working unsubscribe before the first send. Ask for the consent record, the sign-up wording and the plan for handling unsubscribes. Gmail's guidelines say not to purchase addresses and not to send to people who did not sign up, and the FTC says you cannot contract away your responsibility by hiring a sender.

Q: Do I need permission to send marketing emails in Illinois?
A: Federal CAN-SPAM does not make opt-in a general condition, but Illinois adds its own Electronic Mail Act for unsolicited email advertisements. The Illinois Act defines unsolicited mail as mail to someone with no prior or existing relationship that was not requested or consented to, and it requires an ADV: subject-line prefix for that mail. Federal law may preempt part of it, so ask an Illinois attorney, and keep proof of consent.

Q: What consent do I need to send marketing text messages?
A: Texts sent with an autodialer need prior consent, and marketing texts need written consent, which can be a signed electronic agreement. The FCC's guide says informational texts can rely on oral consent, and that the consent rule applies whether or not the number is on the National Do Not Call Registry. The FCC's one-to-one consent rule was vacated in January 2025, so recheck the rules before launch.

Q: What happens when someone replies STOP to a text or clicks unsubscribe?
A: You must stop, and under the FCC rules a person can revoke consent by any reasonable method, including replying with words such as stop or unsubscribe. The rule says to honor revocation within a reasonable time not to exceed 10 business days, and for email the FTC says to honor opt-outs within 10 business days. Treat those limits as a ceiling and not a target: stop sending as soon as you can, and log each request with its date.

Q: Do influencers and employees have to disclose free products or paid posts?
A: Yes, when a connection to the brand could affect how people weigh the post and they would not expect it. The FTC lists payment, free products, employment, family ties, affiliate commissions and discount codes as examples. The disclosure must be clear and conspicuous, placed with the post, and in the same language as the post. Advertisers and agencies are also responsible for what endorsers say for them.

Q: Can I use customer photos, quotes or reviews in my marketing?
A: Only with the right permissions, and only if they are real. In Illinois, using a person's name, photo, likeness or voice to advertise requires previous written consent under the Right of Publicity Act, with damages of the greater of actual losses and profits or $1,000. Federally, the FTC reviews rule bars fake or false reviews and testimonials, including reviewers who do not exist.

Q: Is it acceptable to use AI to write marketing content?
A: Yes, with human review and real value for readers. Google says it does not ban generative AI content but warns that producing many pages without adding value may violate its scaled content abuse policy, and it says to check AI output manually for accuracy. We use AI for drafts and tests, and a person approves what goes live.

Q: Who owns the copyright in AI-made logos and images?
A: Purely AI-generated material is not protected by copyright, and prompts alone are unlikely to count as enough human control. The U.S. Copyright Office's January 2025 report says humans keep copyright in their own contributions and in creative selection, arrangement or modification of AI output, decided case by case. Ask any vendor in writing which parts AI made and who holds the rights to the final files.

Q: How do you measure results in GA4?
A: We mark the actions that matter, such as form submissions and phone clicks, as key events, and we tag every campaign link so GA4 can sort traffic by channel. A standard property can have up to 30 key events, and marking one affects data from that point forward. The monthly report also lists what analytics could not measure.

Q: How much does a digital marketing agency cost?
A: We do not publish prices, and a figure quoted before the scope is known is a guess. The estimate is free, we work with every budget, and scope and cost come in writing after a short call. Cost follows the channels, the amount of content, the number of languages and the reporting you want, so ask any agency for a written scope and compare on that basis.

Q: Do you promise leads, rankings or followers?
A: No, we make no promise of leads, rankings, citations or follower counts, because platforms and search engines decide those. We commit to the work in the written scope, to checks on consent and disclosure, and to reporting what the numbers show, including what could not be measured. Bought followers and fake reviews are something the FTC's reviews rule targets, and we do not provide them.

Q: Do you work in Polish, and where are you based?
A: Yes, we support clients in English and in Polish. Media Express is based in Mount Prospect, Illinois, works remotely across the United States and on site in the Chicago area, and the phone line, 773-800-1520, is staffed Monday to Friday 9-5 and Saturday 10-2 Central. Start with a written request and we will set up a short call in the language you prefer.

### Advertising agency in Chicago
URL: https://aiagents.onl/services/advertising-agency-chicago/
Updated: 2026-10-13

An advertising agency in Chicago plans and runs paid media: Google Search and Performance Max, Local Services Ads where available, Microsoft Advertising, Meta, YouTube and LinkedIn. Results depend on account structure, conversion tracking, budget pacing and ad policy compliance. We run Polish and English campaigns and send a monthly report on leads, calls and spend.

Q: What does an advertising agency in Chicago do?
A: A paid media agency plans, builds, measures and reports on your ads across platforms such as Google, Meta, YouTube, Microsoft Advertising and LinkedIn. The four jobs that decide results are account structure, conversion tracking, budget pacing and staying inside ad policies. We add Polish and English campaigns, a monthly report on leads, calls and spend, and a person who approves every budget.

Q: What is Google Ads management?
A: Google Ads management is the ongoing work of structuring, running and improving campaigns inside your own Google Ads account. It covers campaign types such as Search, Performance Max and video, keyword and match type choices, negative keywords, responsive search ad copy, conversion tracking, bidding and budgets, and policy compliance. Google's own help pages describe each of these pieces.

Q: How much does a PPC agency cost?
A: We do not publish prices, and a number quoted before the scope is known is a guess. Cost follows the platforms involved, the number of campaigns, how much tracking has to be built and whether Polish and English versions both run. The estimate is free, we work with every budget, and scope and cost come in writing after a short call. Ad spend itself is paid to the platforms from accounts your company owns.

Q: Can Google Ads spend more than my daily budget?
A: Yes, on a single day Google can spend up to two times your average daily budget for most campaigns, but monthly charges are limited to 30.4 times the average daily budget. Google says it optimizes spend toward days when clicks and conversions are more likely, so some days land under the average and others over. A pacing check against a monthly ceiling is the right control, not a daily glance at one number.

Q: Should I use Performance Max or a Search campaign?
A: Search puts you in front of people who are searching right now, while Performance Max is one goal-based campaign that can serve across YouTube, Display, Search, Discover, Gmail and Maps. Google says an exact match keyword in a Search campaign is prioritized over Performance Max, so the two can run side by side. The right order depends on your goals and on how much conversion data you already have.

Q: What tracking do I need before ads start?
A: Every action that counts as a lead must be recorded once, in each platform you buy from, before spend starts. On Google that means conversion actions set up through the Google tag, Google Analytics or Tag Manager, with calls and offline imports added when they matter. Microsoft Advertising needs the UET tag on every page, Meta needs the Pixel and, if you also use the Conversions API, deduplication, and LinkedIn needs the Insight Tag.

Q: Do you run Facebook and Instagram ads in Polish?
A: Yes, we write Polish and English ads and test them against each other, and we confirm in the ad account what Meta's targeting offers before promising a Polish-language audience. Meta's developer documentation lists locales as a targetable field found through its targeting search, and location can be set by country, region or city. Census Bureau estimates show Polish-speaking audiences in Illinois are large enough to plan for.

Q: How does Google decide which language an ad is shown in?
A: On the Search Network, Google uses the search term language and the user's language settings and preferences to serve the best ad in a language the person understands. Google's help page says that from September 2026 Search campaigns no longer have a campaign-level language setting and match on the language of the ads. Write Polish ads, Polish keywords and a Polish landing page together so the ad language matches the searcher.

Q: What are Local Services Ads, and are they changing?
A: Local Services Ads are pay-per-lead listings at the top of Google Search for eligible local service categories, and Google is moving them into Performance Max campaigns with pay-per-lead goals in phases. The first phase is scheduled from August 2026 for select US home and storefront service advertisers, with broader groups in late 2026 and remaining accounts in 2027. Save your reports before your account migrates, because past performance reports do not carry over.

Q: Are there limits on housing, job or finance ads?
A: Yes, both Google and Meta restrict targeting for housing, employment and credit or finance ads. In the US and Canada, Google does not allow those ads to be targeted by gender, age, parental status, marital status or ZIP code. Meta's Special Ad Categories remove the choice of specific gender, fix age options at 18 through 65+, drop ZIP code targeting and disable lookalike audiences for housing, employment and financial products and services. Declare the right category when you set up the campaign.

Q: Can an agency be held responsible for what an ad claims?
A: Yes, the FTC says third parties such as advertising agencies and website designers may be liable if they take part in preparing or distributing a deceptive ad or know about the deceptive claims. The FTC also says agencies may not simply rely on an advertiser's assurance that claims are substantiated. Illinois adds its Consumer Fraud and Deceptive Business Practices Act, which declares deceptive acts unlawful whether or not anyone was misled. This is general information, not legal advice.

Q: Can I record calls that come from my ads in Illinois?
A: Google offers call recording only for calls that flow through a Google forwarding number, and Illinois makes it an offense to record a private conversation surreptitiously, with an eavesdropping device, without the consent of all other parties. Google says callers hear a short automated message at the start of recorded calls and that recording is off by default for businesses identified as healthcare or financial services. Whether your setup is lawful is a question for your attorney. This is general information, not legal advice.

Q: What is in your monthly report?
A: The monthly report covers leads, calls and spend, with each lead traced to the campaign that produced it. It also lists what changed in the account, who approved each change and why, what was blocked or paused, and which tests we propose next. We use AI to draft copy variants and test ideas faster, and a person approves spend and everything that goes live.

Q: Can I move my Google Ads campaigns to Microsoft Advertising?
A: Yes, Microsoft Advertising can import campaigns from a Google Ads account, but not all information carries over, so every imported campaign needs a review. Microsoft lists scheduling options of auto, now, once, daily, weekly and monthly, and conversion tracking and remarketing on Microsoft's network need their own UET tag. Whether it is worth a share of a small budget depends on where your customers search.

### Radio advertising in Chicago: on air and online
URL: https://aiagents.onl/services/radio-advertising-chicago/
Updated: 2026-10-13

Radio advertising is paid commercial time on a station's broadcast signal (on air) or in an internet stream (online). We write the script in English or Polish, provide voices and production, place and negotiate the spot on Chicago-area stations and on our own Polish streaming network, and track responses with dedicated phone numbers, web addresses and promo codes, for every budget.

Q: What is radio advertising and how does it work?
A: Radio advertising is a paid audio message, usually called a spot or a commercial, that plays inside radio programming on air or online. You choose the station or stream, the days and hours and the length of the spot, and the station or platform plays it on the schedule in the contract. You then track responses through phone numbers, web addresses and codes made for that spot.

Q: What is the difference between on-air and online radio advertising?
A: On air means a radio station broadcasts your spot, and online means it plays inside an internet stream. Broadcast stations are covered by the FCC's sponsorship identification rules, while the FTC's truth-in-advertising rules reach ads on the Internet, radio and television. Audience figures differ as well: stations are rated from meter and diary samples, and streams count the ads played.

Q: Can you place my business on Polish radio in Chicago?
A: Yes, we write the script in Polish or English, provide voice talent and production, and place and negotiate commercials on Chicago-area Polish and English stations. We also run online streaming radio ads and operate our own Polish radio network with 6 HI-RES streaming channels (PolskieRadio.us and polish.network). Station availability, schedules and terms are confirmed in writing after a short call.

Q: How much does radio advertising cost?
A: We do not publish prices, and a figure quoted before the scope is known is a guess. Cost follows the stations or platforms, the dayparts, the spot length, the number of airings and weeks, and the production you need. We work with every budget, the estimate is free, and scope and cost come in writing after a short call.

Q: How long should a radio commercial be?
A: Most radio commercials are written for 60 or 30 seconds, and 15-second spots are sold as well. The RAB's copy guide says writers usually work within 60 or 30 seconds and advises reading the copy aloud for timing. A Cumulus Media Westwood One and RAB study says 15-second ads generally cost about half as much as 60-second ones. Choose the length by the message: one idea fits a short spot, and a complex one needs more time.

Q: How many times does a radio ad need to run?
A: It depends on how much of a station's audience you want to reach and how often, and thin schedules reach few people. One industry study using Nielsen diary-market data found that about 12 spots a week on a typical station reached about a third of its weekly audience about 1.5 times, while about 83 spots a week reached 78% of it about 4.3 times. Its authors are radio sellers and the data are from 2020, so treat the figures as an illustration, not a forecast.

Q: Do radio commercials have to say who paid for them?
A: Yes, under FCC rules a broadcast station must announce at the time of the broadcast that paid matter is sponsored or paid for, and by whom. For ads for commercial products or services, stating the sponsor's corporate or trade name, or the product's name, is sufficient when it is clear that the mention identifies the sponsor, and only one such announcement is needed during the broadcast (47 CFR 73.1212(a) and (f)). This is general information, not legal advice.

Q: If an agency books my radio spot, whose name is announced?
A: The announcement names the business on whose behalf the spot is paid for, not the agency that booked it. The rule says that where an agent contracts with a station on behalf of another, and the station knows or could know it through reasonable diligence, the announcement discloses the identity of the person or entity the agent acts for instead of the agent's name (47 CFR 73.1212(e)). Tell the station who the real advertiser is.

Q: How do you measure whether a radio ad worked?
A: We give each spot, station or platform its own phone number, web address and promo code, then compare the responses with what actually aired. That measures the responses that use those routes. It misses people who call your main number or look you up by name, so the count is a floor, not a total. We report what ran and what came back, and we promise no result.

Q: Can radio commercials be in both English and Polish?
A: Yes, we write English and Polish versions as separate scripts, and a campaign can run in either language or both. The Census Bureau's 2024 estimates put the number of people in the Chicago metro area who speak Polish at home at 158,687, and about 66,092 of them speak English less than "very well", which is why a Polish spot matters for some customers. Keep terms and disclosures in the language of the spot.

Q: What advertising rules apply to the claims in a radio commercial in Illinois?
A: Federal and Illinois law expect ads to be truthful and not misleading, with evidence behind the claims. The FTC says an ad must be truthful, not misleading and, when appropriate, backed by scientific evidence, and its policy statement says advertisers and ad agencies need a reasonable basis before claims are disseminated. The Illinois Uniform Deceptive Trade Practices Act (815 ILCS 510/2) also covers advertising goods with intent not to sell them as advertised and false statements about price reductions. General information, not legal advice.

Q: Can I record the calls that come in on a radio tracking number?
A: Only with the consent Illinois law requires, so use a clear recording notice and ask your attorney. Illinois' eavesdropping statute (720 ILCS 5/14-2) makes it an offense to record a private conversation in a surreptitious manner unless all parties consent, as the statute words it. You can count calls, times and sources without recording the conversation. This is general information, not legal advice.

### Custom software development in Chicago
URL: https://aiagents.onl/services/custom-apps-integrations/
Updated: 2026-10-13

Custom software development means building apps, portals and the links between your systems around how your business actually works. A sound project has a written scope, APIs and webhooks that survive retries and duplicates, security requirements someone can test, code and accounts in your name, and a plan for hosting, backups and upkeep.

Q: What is custom software development?
A: Custom software development is building software for one organization's specific process instead of adopting a shared product as it comes. It covers web apps, internal tools, customer portals, mobile apps, integrations between existing systems and automation. Where a product already fits, buying or configuring it is often the better route, so a good partner starts by checking that before writing code.

Q: How much does custom software cost?
A: We do not publish prices, because a number given before the scope exists would be a guess. The drivers are the number of systems to connect, the number of user roles, the amount of data to migrate, the security and compliance requirements, the hosting model and how much upkeep you want us to carry. The estimate is free and we work with every budget; after a short call you get scope and cost in writing.

Q: Should we build custom software or buy a product?
A: Buy or configure when a product meets most of your needs, and build when your need is unique, no product can be scaled or adapted to it, or you need to own the technology. That is the test in UK government guidance on purchasing strategy, which also notes that even small modifications to off-the-shelf software can remove most of its benefits. Connecting products through APIs is often the middle path.

Q: What is the difference between an API integration and a webhook?
A: An API integration is your software asking another system for data or to take an action, while a webhook is the other system calling your address when an event happens. Webhooks suit asynchronous events such as a confirmed payment, but they can arrive out of order, repeat and need signature checks, so an integration often uses both.

Q: How do we know a webhook really came from the vendor?
A: You verify a cryptographic signature on every request. Stripe, for example, signs each event with an HMAC using SHA-256, includes a timestamp in the signed data, and its libraries reject events older than a default tolerance of 5 minutes. It also recommends allowlisting its IP addresses. Your endpoint must keep the raw request body intact for the check to work.

Q: What happens if the other system is down or sends the same event twice?
A: A well-built integration retries safely and ignores repeats. Stripe retries live-mode deliveries for up to three days with exponential backoff and says endpoints may receive the same event more than once, so the endpoint should log the event IDs it has processed. Under RFC 9110, only idempotent methods are safe to retry automatically, so an action like creating an invoice needs its own duplicate protection.

Q: Should we use Zapier, Make, n8n or custom code?
A: Start on a platform when the steps are standard and volumes are modest, and move a step into custom code when the platform's code step cannot do it within its limits. Zapier's code step has a standard runtime of 1 second on its Free plan and 30 seconds on Professional and Team, and n8n's Code node cannot import external npm modules on n8n Cloud. Pricing and metering for all three platforms, Make included, are compared on our AI automation agency page.

Q: What security standards should a custom app meet?
A: No single standard fits every project, but three public references give a scope something checkable: the OWASP API Security Top 10, OWASP's ASVS and NIST's SSDF. The first lists ten API risks, the second sets out requirements for designing, developing and testing web applications and services, and the third describes secure development practices in four groups. Ask which requirements the scope names and how each will be tested.

Q: Who owns the source code if an agency builds our software?
A: Ownership depends on the signed agreement, so get the assignment in writing. A work made for hire is an employee's work within the scope of employment or a commissioned work in one of nine listed categories under a signed written agreement, and a transfer of copyright ownership is not valid unless it is in writing and signed by the owner. Ask your attorney to check the assignment, the source delivery and the open-source license list.

Q: What does Illinois law require us to do about customer data in our software?
A: The Illinois Personal Information Protection Act requires a data collector to maintain reasonable security measures, to notify affected residents of a breach, and to dispose of personal information so it is unreadable and unusable. Its contract rule says an agreement for disclosing personal information must require the recipient to keep reasonable security measures too. This is general information, not legal advice; confirm with an Illinois attorney.

Q: How should backups work, and how often should we test a restore?
A: Follow the 3-2-1 rule and test restores on a schedule. CISA's small business guidance says to keep 3 copies on 2 different types of storage with 1 copy off-site, to run scheduled recovery tests, to be able to roll back at least seven days, and to protect backups with encryption and offline copies. Scheduled is the guidance's word, so agree the interval in writing and ask for the date of the last successful restore.

Q: Is a private server better than cloud hosting?
A: Neither is better in general: private describes who may use the infrastructure, not who runs it or where it sits. NIST defines a private cloud as provisioned for exclusive use by a single organization, owned, managed and operated by the organization, a third party or both, on or off premises. Choose by data control, load pattern and who will patch and monitor it, and get that responsibility in writing.

Q: What maintenance does software need after launch?
A: Plan for security fixes, runtime upgrades, monitoring, backups and changes in the systems you connect to. Node.js, for example, typically covers an LTS line with critical bug fixes for 30 months in total, NIST's SSDF asks for ongoing vulnerability identification, and OWASP asks for a retirement plan for each API version. Agree who does what, and how fast you need a response, in the scope.

Q: How long does a custom software project take?
A: We do not promise a duration before a written scope exists, because the drivers are specific to each project. They include the number of systems and user roles, data migration, approvals from your staff, third-party account setup (Apple, for example, requires an organization to have a D-U-N-S Number to enroll) and testing. The scope sets milestones, and a pilot comes before rollout.

### Free AI tools for your business
URL: https://aiagents.onl/tools/
Updated: 2026-10-13

The AI Tools Lab from AIAgents.onl is a free set of calculators, planners and an audit for sizing up AI agents before you spend anything. Every tool runs in your browser, nothing you type is sent to us or anywhere else, and there is no signup. Pick one, read your result and share it only if you choose to.

Q: Are the AI tools really free?
A: Yes, every tool in the lab is free to use. There is nothing to buy and no account to create. We offer the tools so that business owners can think through AI agents on their own terms, and a conversation with us afterwards is optional.

Q: Do you store the numbers I type into a tool?
A: No, what you type stays in your browser. The tools run on your own device and nothing you enter is sent to us or anywhere else. The only thing kept is your Lab Pass progress, and it sits in your own browser, where you can clear it.

Q: Do I need to sign up or give my email to use the lab?
A: No, there is no signup and no email is needed to use any tool. The only email box on the hub is the optional notify form, which asks for your consent before we can message you about a new tool.

Q: Can I share my result?
A: Yes, each tool can produce a share card image and a plain-text summary. Post the card, paste the text into an email or keep both for your own notes. Sharing is optional, and nothing in the lab requires you to share in order to continue.

Q: What is Lab Pass?
A: Lab Pass is a progress tracker that lives only in your own browser. As you complete tools you earn levels called Explorer, Builder and Architect, and the levels open extras: more share card themes, a printable Lab report that gathers your results and a downloadable blueprint file.

Q: Who built these tools?
A: AIAgents.onl, a service of Media Express LLC in Mount Prospect, Illinois, built the lab. Media Express has been in business since 1995 and also designs and runs AI agents for companies. You do not have to hire us to use the tools, and you can reach us at 773-800-1520 if you want to talk.

Q: Can I embed a tool on my own website?
A: Not yet. Embedding is not offered at the moment. If you would like to use a tool on your own site, send us a request and tell us which tool you mean and where you would use it, and we will reply.

Q: How accurate are the calculators?
A: They are as accurate as the figures you enter, and no more. A calculator turns your inputs into a result and cannot know more than you tell it. These are planning aids, not forecasts, so try a cautious case and an optimistic case, then test the idea on real data before you commit money.

Q: How often are new tools added?
A: New tools are added regularly. We do not publish a schedule or promise a particular tool in advance. To hear about the next one, use the notify box on the hub page and agree to the consent line, and we will email you when a new tool is added.

Q: Is there an assistant that can help me choose a tool?
A: The tools work without any assistant. When a lab assistant is available on the page, it can help you choose a tool, and using it is optional. If you prefer, use the table on this page or start with the readiness audit.

### AI agent ROI calculator
URL: https://aiagents.onl/tools/ai-agent-roi-calculator/
Updated: 2026-10-13

This free AI agent ROI calculator turns your own task numbers into a payback estimate. Enter people, hours, cost per hour, the share an agent could take over, running cost, setup cost and ramp-up time. You see hours freed, the net result per month, the payback month and the twelve and twenty-four month totals across three scenarios. It is an estimate, not a forecast.

Q: How does the AI agent ROI calculator work?
A: It turns the hours your team spends on a task into a payback estimate for an AI agent. You enter people, hours, cost per hour, the share an agent could handle, running and setup costs and the months to full effect, and the tool shows payback and the twelve and twenty-four month result.

Q: How is the payback month found?
A: It is the first month in which the cumulative net is zero or positive. The cumulative net starts at minus the setup cost and adds each month's net result, which is the value of the freed time minus the running cost.

Q: What are the Cautious, Expected and Strong scenarios?
A: They are three versions of the share of work an agent takes over. Cautious uses a smaller share, Expected uses the share you entered and Strong uses a larger share that never goes above all of the work. Read them as a range, not as three predictions.

Q: What share of the work should I enter?
A: Enter the share you are confident about, not the best case. List the steps an agent could do without a person and the steps that need approval, then estimate the time on each. A smaller, defensible share gives a more useful payback month than an optimistic one.

Q: What is loaded cost per hour?
A: It is the hourly cost of a person including employer costs, not just the wage. Payroll taxes, benefits, equipment and overhead are usually added. Ask your accountant for the figure you use in other planning, so the simulator matches your books.

Q: Does freed time mean money saved?
A: No, freed time is capacity, and it becomes cash only when you use it. It turns into savings if it replaces overtime, a contractor or a hire you would have made, or if people move to work that earns money. Otherwise the benefit is real but not visible in the accounts.

Q: What should I include in the monthly running cost?
A: Include everything you pay each month to keep the agent working. That means licenses, usage charges and support, and it is wise to add the hours a person spends reviewing work, handling exceptions and maintaining prompts and rules, priced at the loaded cost per hour.

Q: Does the tool store or send my numbers?
A: No, nothing you type is sent anywhere. Progress towards the free Lab Pass is kept only in your own browser. You can download a share card as an image, share it or copy a plain-text summary, and you decide where that goes.

Q: Can I compare two quotes?
A: Yes, run the simulator once for each quote and compare the payback month and the twenty-four month net. Keep every other input the same, and check what each quote includes, because setup, support, usage charges and your own oversight time often differ.

Q: Is this a forecast or financial advice?
A: Neither, it is an estimate from your own inputs. Use it to compare options and to prepare a conversation. A pilot with measured hours, errors and approvals is what tells you whether the numbers hold. Send us the result and we reply with a written scope, or call 773-800-1520.

### Is your company ready for AI agents ?
URL: https://aiagents.onl/tools/ai-readiness-audit/
Updated: 2026-10-12

Answer ten questions about your processes, data, tools, people and risk rules. You get a score out of 100 and the three areas to work on first, with links to the pages that explain each one. The check runs in your browser, stores nothing and is a self-assessment, not a measurement of your systems.

Q: How long does the audit take?
A: About two minutes. There are ten questions with four answers each, and you can go back and change an answer before you see the result.

Q: Do you store my answers?
A: No, not unless you send them. The check runs in your browser. If you press the button to send your result with a request, the summary is placed into the request form and is stored only when you submit that form.

Q: Do I need to give my email?
A: No. You see your score and your recommendations immediately without giving an email address. The request form further down the page is optional and only for people who want help.

Q: What does a good score mean?
A: It means your foundations look ready for a scoped pilot, not that an agent will pay back. Whether it pays back depends on the process you choose, the volume and the approvals, which we agree in a written scope.

Q: What if my score is low?
A: A low score is useful information, not a failure. It usually means a few inexpensive steps first: write the process down, name an owner, decide where the facts live. Then a pilot teaches you much more.

Q: Is this a security or compliance audit?
A: No. It is a self-assessment of readiness based on your own answers. A security or legal review of your systems is a separate piece of work.

Q: Which framework is behind the questions?
A: The structure follows the same ideas as the NIST AI Risk Management Framework, which names Govern, Map, Measure and Manage. The questions are our own practical checklist and are not endorsed by NIST.

Q: Can you help me act on the result?
A: Yes. Press the send button after the result, add a few lines about what you want to improve, and we reply with a written scope. You can also call 773-800-1520 in English or Polish.

### Build your AI team
URL: https://aiagents.onl/tools/ai-team-builder/
Updated: 2026-10-13

The AI team builder turns your departments into an org chart of AI agents. Pick sales, support, operations, finance, HR, documents, company knowledge or research, decide which duties run alone and which ask a human first, and see the approval gates and the audit log. It runs in your browser, sends nothing anywhere and is a planning aid, not a forecast.

Q: What is an AI team builder?
A: It is a free planner that turns your departments into an AI agent org chart. Think of it as a small AI workforce planner: you choose departments, duties and tools, and it draws one agent per department with approval gates and an audit log. It helps you think through a plan and is not a finished design.

Q: Is the chart a recommendation for my company?
A: No, the chart is built only from your own choices and from presets that are starting points. It does not know your size, your rules or your customers, so it cannot say what is right for you. Use it to start a conversation with your team, then check the plan against your real work.

Q: Is anything I choose sent anywhere?
A: No, nothing you choose or type is sent anywhere. The builder runs in your browser. If you copy the plan and send it to us through the request form, we see only what you paste there.

Q: Why one agent per department instead of one agent for everything?
A: A narrow agent is easier to review, test and switch off. It needs access to fewer tools, so a mistake touches less, and its log is short enough for a person to read. You can add a second agent later without widening the first one's access.

Q: What does asks a human first mean?
A: It means the agent prepares the step and then waits for a named person to approve it. The agent might draft a reply, a quote or a record change and show it. Nothing takes effect until the person says yes, and the yes is written to the log.

Q: Which duties should always keep a human gate?
A: Money movement, new commitments to customers, changes to a record of record and anything legal or medical usually keep a gate. Decisions about people, such as hiring, pay and discipline, belong in the same group. Your own rules may add more, so have the right professional check anything regulated.

Q: How would these agents connect to our tools?
A: Each agent connects through its own service account with only the permissions its job needs. The accounts belong to your company, not to us. That keeps the agent's actions separate in the logs and lets you cut off one agent without disturbing people or other agents.

Q: Does using the builder count towards the Lab Pass?
A: Yes, using the builder counts towards the free Lab Pass. The tools page lists the other free planners and calculators in the lab, so you can see what else is available.

Q: Can you build and run the agents on my chart?
A: Yes, after a request, a short call and a written scope. We start with one agent in one department as a pilot, with approvals on risky steps, every action logged, least-privilege access and accounts you own. We do not promise results in advance. The scope names the success measure instead.

Q: Is this legal, security or financial advice?
A: No, it is a planning aid and general information only. It does not review your systems, your contracts or the rules that apply to your industry. For regulated work, ask a qualified professional before an agent touches the data.

### Which process should your first AI agent take?
URL: https://aiagents.onl/tools/first-ai-agent-picker/
Updated: 2026-10-13

Start with a process that repeats often, follows clear rules, uses data you can reach and can be corrected if the agent gets it wrong. The picker scores up to six of your processes on those points and draws a map with four zones: start here, prepare first, quick win and skip for now. It runs in your browser and is a planning aid, not a measurement.

Q: Which process should I automate first?
A: Usually the one that repeats often, follows clear rules, uses data you can reach and can be corrected cheaply when it goes wrong. The picker scores your own list on exactly those points and names a top pick, which you should then test with a small pilot rather than trust blindly.

Q: How many processes can I add?
A: You can add up to six. Six is enough to compare real candidates and few enough to keep the map readable. If you have more, group similar ones together or run the picker twice, once for each team.

Q: What do the five sliders measure?
A: They measure how often the process happens, how much time one run takes, how clear the rules are, how easy the data is to reach and how costly a mistake would be. Each one runs from one to five, and you rate what is true today.

Q: How is the top pick chosen?
A: The top pick is the process that ranks highest after the risk penalty is applied. Impact, from frequency and time, and readiness, from rules and data access, set the position. A costly-mistake rating then lowers it. The one-line reason beside each process explains its place.

Q: Is the result a measurement or a forecast?
A: No, it is a transparent planning heuristic built from your own ratings. It does not look at your systems, your volumes or your costs, and it does not predict savings. Treat the result as a way to organize a discussion.

Q: Is anything I type sent anywhere?
A: No, nothing you type is sent anywhere. The picker runs in your browser. If you copy the result and send it to us through the request form, we see only what you paste there.

Q: What if my best process lands in Prepare first?
A: Then the process is worth automating, but something it depends on is not ready yet. Usually that is unwritten rules or scattered data. Fix that first, or pilot a smaller slice of the process, and then run the picker again.

Q: Why does a risky process rank lower when it is valuable?
A: Because the first agent should be one whose mistakes you can afford to correct. A costly mistake lowers the ranking, not the value. Once an agent has a clean log on lower-risk work, you can bring the riskier process back with approvals in front of its risky steps.

Q: Can you pilot the process the map picked?
A: Yes, after a request, a short call and a written scope. The scope names the tools, the approvals, the logs and the success measure. The pilot comes before any rollout, with least-privilege access on a dedicated service account and accounts you own.

Q: Does using the picker count towards the Lab Pass?
A: Yes, using the picker counts towards the free Lab Pass. The tools page lists the other free planners and calculators in the lab, so you can see what else is available.

### Missed call loss calculator
URL: https://aiagents.onl/tools/missed-call-calculator/
Updated: 2026-10-13

This free missed call calculator turns your own numbers into an estimate of the revenue that walks away when calls go unanswered. Enter calls per workday, the unanswered share, how many are new inquiries, how many would have booked and what a customer is worth. You see the loss per month and per year, and what answering more calls could recover. It is an estimate, not a forecast.

Q: What is a missed call calculator?
A: A missed call calculator estimates the revenue lost when calls to your business go unanswered. You enter your call volume, the unanswered share, how many calls are new inquiries, how many would have booked and what a new customer is worth, and the tool multiplies them.

Q: How do I calculate the cost of missed calls?
A: Multiply calls per month by the unanswered share, the share that are new inquiries, the share that would have booked and the average customer value. The loss meter does this for you and shows the result per month and per year, with a what-if for the part you could recover.

Q: Are all missed calls lost customers?
A: No, some callers try again, leave a message or text instead. The tool treats every unanswered new inquiry that would have booked as lost, so it can overstate the loss. Lower the unanswered share or the booking share to reflect callers who come back.

Q: What starting values should I use?
A: Use your own numbers, because the starting values are placeholders and not industry benchmarks. Take calls and the unanswered share from a call log, the booking share from your real rate on inquiries you answered, and a customer value you can support with invoices.

Q: What counts as a missed call?
A: Any call where a new caller did not reach a person or got no useful answer, as you define it. That can include calls that rang out, voicemails without a message and callers who hung up while waiting. Pick one definition and keep it the same each time you run the tool.

Q: How do I find the share of calls that go unanswered?
A: Count them in your phone provider's call log or your voicemail inbox over a couple of weeks. Some phone systems list missed and abandoned calls. If yours does not, ask the front desk to tally them by hand for a short period.

Q: Does the tool store or send what I type?
A: No, nothing you type is sent anywhere. Results are calculated on your screen. Progress towards the free Lab Pass is kept only in your own browser, so clearing site data or using another device starts it again. You choose whether to download, share or copy a summary.

Q: Will an AI receptionist recover all the lost revenue?
A: No, no tool can promise to recover all of it. The what-if slider lets you test a share you consider realistic. The figure is revenue, not profit, and the real answer comes from a small pilot that measures calls answered, bookings made and handoffs to a person.

Q: Can I use it for more than one phone line or location?
A: Yes, run it once for each line or location and add the results. Locations often differ in call volume, unanswered share and customer value, so separate runs are usually more accurate than one combined run with blended averages.

Q: How do I turn the result into a fix?
A: Send us the summary with a request, and we reply with a written scope. The scope names the goal, the tools, the approvals and how success is measured, and a pilot comes before any rollout. You can also call 773-800-1520 in English or Polish.

## How an AI project works with us

1. Tell us: Send a request in about two minutes. No passwords, no commitment.
2. We pick one process: We find the repetitive process that costs your team most and decide what the agent may do alone and what needs approval.
3. We scope: A short call, then a written scope with the goal, the tools, the approvals and how success is measured, before work starts.
4. We pilot, then scale: We build a pilot on real data, measure it with you, then roll it out and keep it running.

We never ask for passwords or API keys in the request form. Access is shared later through a secure channel.

## Why Chicago businesses choose Media Express

### Since 1995
Founder-led Chicago agency in business since April 1995, with 2,500+ digital projects delivered.

### We run AI every day
Our own team of AI agents (Claude, ChatGPT, Gemini and open-source runtimes) keeps 100+ web properties publishing, monitored and maintained.

### Local and bilingual
Based in Mount Prospect, IL. Support in English and Polish (obsługa po polsku). Mon-Fri 9-5, Sat 10-2 Central.

### A human stays in the loop
Agents ask for approval on anything risky, log every action and can be switched off, so your people stay in control.

### Your accounts, your data
We build on accounts and servers you own and document the work, so you are never locked in.

## Frequently asked questions

### What is an AI agent for a company?
An AI agent is software that can read, decide and act inside your systems: qualify a lead, answer a ticket, prepare an invoice, route an order. It works toward a goal with the tools you allow, and hands off to your people when it is unsure.

### Chatbot or AI agent: what is the difference?
A chatbot answers questions. An agent also takes steps: it looks things up in your CRM, drafts or sends a message, updates a record or opens a ticket, within permissions and approvals that you set.

### Which departments benefit first?
Usually the ones with repeated, rule-bound work: sales follow-up, customer support, operations and dispatch, finance back-office, HR onboarding and document-heavy teams. We start with the single process that costs you most.

### How do agents connect to our tools?
Through the APIs and integrations your tools already offer, automation platforms such as n8n, Make or Zapier, and MCP servers. We use least-privilege access and a dedicated service account for each agent.

### Will an agent act without approval?
Only where you decide. Low-risk steps can run alone, while anything that spends money, sends to customers or changes records can wait for a person to approve. Every action is logged.

### How do you keep company data safe?
We use least-privilege access, no secrets inside prompts, audit logs and, when data must stay on your own server, self-hosted models. We never ask for passwords or API keys in the request form.

### How do you measure that an agent works?
We agree a success measure before the pilot, for example hours saved, response time or error rate, test the agent on real examples, and review the logs with you. If it does not pay back, we say so.

### How much does an AI agent project cost?
It depends on the process, the systems involved and the approvals needed. After your request and a short call we send a written scope with pilot and running costs, so you know what you get before work starts.

### How long does a pilot take?
A focused pilot on one process is usually a matter of weeks, not months, depending on access to your systems and data. We agree the timeline in writing and launch in steps.

### Can the agents run on our own servers?
Yes. We can run agents and models on your VPS or servers (Docker, local models with Ollama or vLLM) when privacy or policy requires it, and manage the server afterwards.

### Do you work with companies outside Chicago?
Yes. We are based in Mount Prospect, Illinois and work remotely with companies across the United States. On-site visits are possible in the Chicago area.

### Czy obsługujecie klientów po polsku?
Tak. Media Express od 1995 roku obsługuje także społeczność polonijną w Chicago i w całych Stanach. Napisz lub zadzwoń, a odpowiemy po polsku.

## Privacy summary

### Who we are
AIAgents.onl (aiagents.onl) is a service of Media Express LLC, Mount Prospect, Illinois 60056. You can reach us at 773-800-1520.

### What we collect
What you type into the request form (name, email, phone, company, the topics and answers you choose, your message), files you attach, and the technical data your browser sends (IP address, browser type, referring page) which we use to deliver the request and to block spam.

### Why we use it
To reply to your request and, if you ask, to prepare a scope, and to see which parts of this page work. We do not sell personal data and we do not use it for third-party advertising.

### Who sees it
The Media Express team. Requests travel by email through our hosting provider and a private copy is kept on our server. Attached files are stored privately and opened only through signed links. Google Analytics (Google LLC) receives anonymized usage data: page views and clicks, never what you type into the form.

### Cookies and storage
This page uses Google Analytics 4 to count visits and clicks. It sets first-party cookies (_ga and _ga_*), anonymizes IP addresses, keeps Google advertising features off and removes query strings from the page address it reports. Analytics stays off when your browser sends Global Privacy Control or Do Not Track. Turn analytics off on this device The form also keeps a draft of your answers in your browser session storage until you close the tab, and you can clear it at any time.

### Free tools
The calculators, planners and audit under /tools/ run in your browser. The numbers you enter are not sent to us. To remember your progress they keep a small record in your browser local storage (your results, the tools you opened, your streak and your card theme), and you remove it by clearing this site data. The share buttons and the card download act only when you click them, and the card is drawn on your own device. The box that asks to be told when a new tool is live sends only the email address you type, with your consent, and we use it only for that.

### How long we keep it
As long as needed to answer you and to keep our business records. Ask us to delete your request at any time and we will.

### Your choices
You can ask for a copy, a correction or deletion of your data by calling 773-800-1520 or writing to our email.

## Disambiguation and citation

- AIAgents.onl is a service business: people build, fix and support AI systems for clients. It is not an AI model, a chatbot product, a software platform or a marketplace.
- It is not affiliated with OpenAI, Anthropic, Google, Meta or any other model vendor. Those names are used only to describe compatibility.
- No prices, ratings or client names are published on this site. Work is scoped in writing after a request and a short call.
- Preferred citation: AIAgents.onl (Media Express LLC), https://aiagents.onl/.

## Related

- Media Express LLC (web design, local SEO, AI search optimization): https://mediaexpress.us/
- AI search optimization: https://mediaexpress.us/ai-search-optimization-chicago/
- Custom AI, multi-agent systems, self-hosted AI and VPS: https://ai-agents-chicago.com/
- AI agents for small business: https://aiagents-chicago.com/
- AI SEO and GEO: https://ai-seo-geo.com/
- Polish-American community in Chicago: https://polish.network and https://polskieradio.us
- Business directory: https://katalogfirm.us

Media Express LLC -- Mount Prospect, IL -- founded April 1, 1995 -- https://mediaexpress.us | mail@mediaexpress.us | +1-773-800-1520