What the picker does
Most companies have more candidates for automation than patience. The hard part is not finding processes an AI agent could touch. It is deciding which process to automate first with AI, so that the first agent has a fair chance to work and the lessons carry over to the second.
The picker is a small piece of AI use case prioritization. You list up to six processes, each with a short name, and move five sliders for each one. It turns your ratings into two scores, draws an animated bubble map, lists the processes in order with a one-line reason for each and names the top pick.
How to use the picker
It takes about three minutes and needs no signup.
Name your processes
Add up to six, in a few words each, such as quote follow-up, invoice matching or new-hire paperwork. Use the names your team already uses.
Rate five things
For each process, rate from one to five how often it 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. Rate what is true today, not what you hope for.
Read the map
Each process appears as a bubble in one of four zones: start here, prepare first, quick win or skip for now.
Read the ranking
The list under the map gives the order, a one-line reason for each process and the name of the top pick.
Copy or share
Copy the result as text or save a share card. Using the picker also counts towards the free Lab Pass.
If you are unsure about a rating, pick the more cautious one. A picker that errs on the side of caution sends you to a pilot you can finish rather than one that sounds impressive.
How the scores work
The scoring is simple on purpose, so you can argue with it. Three ideas drive the result.
Impact
Built from how often the process happens and how much time one run takes. A process that happens often and takes long has the most time to give back.
Readiness
Built from how clear the rules are and how easy the data is to reach. A process with written rules and its facts in one reachable place is ready. One that runs on judgment and scattered files is not.
Risk penalty
How costly a mistake would be lowers the ranking. The more damage one wrong step could do, the lower the process falls, because the first agent should be one whose errors you can afford to correct.
Impact and readiness decide where a bubble sits on the map. The risk penalty works on the ranking, so a valuable process with costly mistakes can rank lower in the list than its zone alone would suggest.
How to read the four zones
The map plots two things for every process: its impact and its readiness. The four zones are the four corners of that picture.
| Zone | What it means | What to do |
|---|---|---|
| Start here | High impact and high readiness | Take it to a pilot scope. It repeats often, the rules are clear and the data is within reach. |
| Prepare first | High impact but low readiness | Worth automating, but something it depends on is missing, usually unwritten rules or scattered data. Fix that first. |
| Quick win | Lower impact but high readiness | Easy to automate and small. A good way to build confidence, or a candidate for the second agent. |
| Skip for now | Low impact and low readiness | Leave it alone. Come back if the volume grows or the process changes. |
A bubble close to a border is a prompt to talk, not a verdict. Change one rating and run it again. If a single small change moves a process to another zone, your answer was uncertain, and that uncertainty is worth discussing with the people who do the work.
What a good first process looks like
When people ask what to automate first, the answer is rarely the most exciting process. It is the one that scores well on five plain points.
- It is frequent. It happens often enough that saved time adds up and the agent sees plenty of real cases.
- It follows rules. A new employee could do it from a checklist, and the exceptions are known.
- The data is within reach. The facts it needs live in a system the agent can be given a scoped account for, not in someone's head.
- Mistakes are recoverable. A wrong step can be spotted and undone without harm to a customer, a payment or a legal position.
- A person can review it. Someone who knows the job can check the agent's work in a few minutes and approve it.
Typical candidates are answering the same customer questions, filing incoming documents, booking appointments and drafting routine follow-ups. Poor first candidates are pricing decisions, hiring decisions and anything where the first run already touches money or a legal commitment.
A first agent that drafts while a person approves is a better start than a first agent that acts alone. You learn how it behaves, and the log shows you where to loosen or tighten.
What to do with a Prepare-first process
A process in this zone is the most common result, and it is good news. It matters, and the gap is something you can close. Work through these steps, then run the picker again.
- Write the process down as steps, including the exceptions people handle from memory.
- Decide where the facts live and who keeps them current: prices, policies, customer history.
- Name an owner on your side and the person who will review the agent's work.
- List the steps that need a person's approval before an agent acts.
- Re-rate the process and see whether it moved towards the start-here zone.
Often the answer is a smaller slice: the most rule-bound part of the process, where the data is already reachable. The AI readiness audit shows where the gaps are across your company, and AI consulting covers the planning work around them.
When this is not a good fit
The picker is a thinking aid. It helps least in these cases.
- You have only one process in mind. There is nothing to compare. Go straight to a short call and a scope.
- You need a financial case. The picker ranks processes. It does not estimate cost or payback. The AI agent ROI calculator is a separate planning aid for that.
- The process is regulated and no rules exist for AI yet. Write the rules first, with the right professional, before any agent touches the data.
- You want a design. The map picks a place to start. It is not an architecture, a security review or a legal opinion.
In any of these cases, a short call is usually a better first step than another chart.
What to prepare for a pilot scope
Once the map has named a process, the next step is a small pilot, not a full rollout. Bring the following and the scope is quick to write.
- The result you copied from the picker, with the top pick and its ratings.
- A rough description of the process: who starts it, what the steps are and where it ends.
- The tools it touches and who owns each account.
- The steps that must be approved by a person, and the name of the approver.
- A note on sensitive data in the process: health, finance, children, voice recordings, personal files.
- One named owner on your side with time for a pilot.
Request
Send the process the map picked with a few lines about it. The request form is on the home page and at the bottom of this page.
Short call
We go through the tools, the limits and the approvals, and tell you honestly whether an agent fits.
Written scope
The goal, the tools, the approvals and the success measure are written down before anything is built.
Pilot before rollout
One process, with approvals on risky steps, every action logged, least-privilege access on a dedicated service account and accounts you own. You can switch the agent off at any time.
For help with building and connecting the agent, see our AI automation agency page. You can also call 773-800-1520 in English or Polish, Monday to Friday from nine to five and Saturday from ten to two, Central time.
Questions answered
Which process should I automate first?
How many processes can I add?
What do the five sliders measure?
How is the top pick chosen?
Is the result a measurement or a forecast?
Is anything I type sent anywhere?
What if my best process lands in Prepare first?
Why does a risky process rank lower when it is valuable?
Can you pilot the process the map picked?
Does using the picker count towards the Lab Pass?
Pilot the first process
Send us the process the map picked. We reply with a written scope for a small pilot with approvals and logs.