Tools and score
Tools and score First AI agent finderEdition of November 13, 2026 · Proof 1 · Page V1
First AI agent finder: name the work that keeps coming back to you, get the agent that takes it and its four-week pilot.
Three questions, one minute. You pick the work, how often it comes back and who can check the drafts; the finder returns one agent, the plan of its first four weeks and the list of what never leaves your hands.
Get my score in 90 secondsFree, with no account. Loïc Guyon, the founder, reads every result and calls the firms he can help.
Your answers, read on this page.
Continued inQuestions owners ask
Before you decide
A first agent takes one piece of work
The first agent does not transform the whole firm. It takes work that comes back, that a person can check in a few minutes, and whose before and after can be measured. That narrow scope builds trust, because the team sees the gain before any wider access opens.
The work picks the agent
Opportunities that go quiet go to Bob. Approved work waiting for a handoff goes to Eva. Enquiries waiting for a reply go to Nestor. Scattered context goes to Memo. Invoicing and collection go to Moni. They are the Scorecard’s five agents, each tied to the leak it measures.
Four weeks to decide
Week 1, the inputs: the agent reads the approved sources. Week 2, the quality: each draft is compared with what you would have done. Week 3, the approval rules. Week 4, the measure: is the work faster, with no loss of quality? You then widen it, narrow it or stop.
Questions owners ask
Which AI agent should I build first?
The one that takes your most frequent work with the simplest review. In many firms that is business development follow-up, replies to new enquiries or meeting preparation.
Should my first AI agent connect to every tool?
No. It connects to the few sources it needs to prove the work. Every extra integration slows the pilot and makes the review harder.
How long should the first pilot run?
Four weeks is enough for narrow work. For monthly work, the pilot runs on five past files, so you compare without waiting for the next case.
Who should own the first agent internally?
The person closest to the work, who can judge an output. They approve the drafts and say whether the agent makes the work easier.
How is this different from the Scorecard?
The Scorecard measures where the whole firm loses time, in twelve questions. This finder starts from one piece of work you name and returns its pilot. They work together: start with the Scorecard if you do not know where to begin.
Next, see where the whole firm stands.
Ninety seconds on how your firm runs today. You see where senior time goes, and when twenty minutes on one engagement would help, Loïc calls you.

AI Jungle LLC, run by Loïc Guyon.

