The method
The Leverage Loop: diagnose, test, build, operate, improve.
One method for every AI Jungle engagement, run on one process of your firm at a time. Each step hands you something you can use, names who decides and says how long it takes.
The brief is free, with no account, and Loïc reads it before the call.
Every step leaves something the firm can use.
01
Diagnose
A written report to decide on
- What happens
- Two 45-minute sessions with Loïc Guyon on one process of your firm, and a report between them: a reasoned recommendation, the alternatives, including no AI, and an action plan reviewed by a human consultant.
- Who decides
- You. At the second session you choose to test, take a simpler fix, wait or stop.
- How long
- About two weeks, from the first session to the second.
- Entry point
- The Leverage Assessment, from $999 excluding tax, deducted in full if you continue. See the Assessment
02
Test
Your own files, scored
- What happens
- Before any build, the work runs on five anonymised files from your firm. Your reviewer compares each draft with the version they would have approved and marks what is right, missed or wrong.
- Who decides
- Your reviewer scores the drafts. On those scores your sponsor decides on the build, and can stop the trial at any time.
- How long
- Two weeks, starting after the second session, on five files agreed with your sponsor.
- Entry point
- The Leverage Assessment’s action plan, which sets the trial, its owner and its stop conditions. See a sample report
03
Build
One tested process, in your tools
- What happens
- We configure the sources, rules, permissions and approval points in your firm’s own installation, then test normal cases and exceptions with your team. Each week, your corrections shape the process before it takes on more.
- Who decides
- Your reviewer checks the outputs against criteria agreed before the build. You accept a bounded release, ask for corrections or keep the existing process.
- How long
- About 30 days, while your current mandates carry on.
- Entry point
- The Build, $15,000 once, or waived with the Flash option for a 12-month commitment, two personal introductions and a signed case study at month 6. Read the terms
04
Operate
A running process, and its record
- What happens
- We maintain the integrations and review outputs, incidents and corrections with your operator. Everything runs in an installation dedicated to your firm, hosted in Germany, and your data never trains a model.
- Who decides
- Your team approves every message before it goes out and can restrict or suspend an action while an exception is resolved.
- How long
- Ongoing: weekly iterations and a monthly review.
- Training
- A named colleague handles a representative case and explains the decision back, until your team runs the process without us.
- Entry point
- The Leverage Program, the same for every firm: $5,000 per month for the whole firm, plus an alignment share, never more than $15,000 in a month. See the Leverage Program
05
Improve
A measured gain, then the next process
- What happens
- Each correction your team approves becomes part of the working method, so the next draft starts closer to what you would have written. We measure the indicator chosen at the Assessment against its starting point.
- Who decides
- You. Keep improving, extend to the next process, or take the work back with the handover pack.
- How long
- At every monthly review, with a first checkpoint at 90 days.
- Handover
- The handover pack stays current throughout, with documentation, exports, configuration and open issues. The exit is written in the first conversation.
- Guarantee
- Under the Leverage Program, if the indicator chosen at the Assessment has not improved by at least 10% at 90 days and your side of the conditions is met, you leave without paying the build, and you keep the memory, data and tools listed in the handover pack.
- Entry point
- The same engagement as Operate, the Leverage Program. Understand the handover
What Improve teaches feeds the next Diagnose, so the second process starts further ahead than the first.
Choose who will carry the work.
These are operating choices, not a ranking of suppliers. The right choice depends on who can own the process after the first delivery.
Scroll the table sideways to see all four columns.
| Mode | Who runs it | When it fits | What to settle first |
|---|---|---|---|
| Keep the process manual | Your team | The work is rare, changing or mostly judgment. | Document the handover and the decision owner. An AI build may add no value. |
| Use software directly | Your team, with product support | You have an operator who can configure, review and maintain it. | Check the product’s permissions, export and licence terms. |
| Commission a defined build | Your team after the agreed handover | The process is stable and you can take over its operation. | Specify acceptance tests, documentation and the support boundary. |
| Choose a managed partnership | AI Jungle operates the agreed scope; your team approves business decisions | You need continuing maintenance, training and improvement. | Name the operator, reviewer, review cadence and contractual exit terms. |
The exit belongs in the first conversation.
Before starting, put the operating boundary in writing. The paid proposal or order form defines the scope and the rights attached to each deliverable.
- Access
- Name the accounts, source material and authorised people. Define which actions need approval and who can withdraw access.
- Stopping a process
- Define the stop conditions, the person who receives the exception and the manual fallback. A missing source or approval is a reason to stop and ask.
- Restitution
- List the documentation, data exports, configurations and operating records included in the handover, their formats and any open issues. Agree the receiving operator and the transfer of credentials separately.
- Rights and dependencies
- Specify ownership and usage rights in the agreement. Third-party software and licences keep their own terms; this is not an automatic transfer of every tool or component.
An assessment does not commit you to a Build or a managed partnership. Standard has no 12-month commitment.
Flash: the build is waived for a 12-month commitment, 2 personal introductions to leaders of qualified firms and a signed case study at month 6. At day 90, if you are not satisfied and the targets are not met, you stop with no fees, no build to repay and no other condition.
Leaving the managed partnership means the documented handover agreed at the start; third-party services keep their own terms.
AI Jungle · Capability brief
Make the firm’s know-how survive the handover.
Operational AI consulting for owner-led B2B service firms. Before a sale, after an acquisition and throughout growth.
The print view contains this brief only. You can save it as a PDF from your browser.
- For a seller
- Expose the work that depends on the owner. Document a priority process and test whether another person can carry it forward.
- For a buyer
- Examine operational continuity, identify unresolved dependencies and build a measured post-acquisition improvement plan.
- For an adviser
- Add an operational workstream to the transaction. Agree the mandate, authorised information and deliverables shared with each party.
What we can work on
Knowledge capture and handover · AI-supported proposals, preparation and follow-up · operating systems and team training · digital and media workflows · a scoped test for international development.
What the engagement makes explicit
Recommendation and alternatives; authorised sources; a bounded process; human review; acceptance and stop criteria; documentation and agreed handover. Each engagement selects the relevant scope.
Our role in a transaction
We support operational readiness and implementation. The sale, the valuation and the return also depend on the parties and the market, and legal, tax, financial and transaction advice remains with the appointed professionals. Separate mandates and permissions are needed to work with different parties.
Inspect the work
The Asia-Connect case shows an operational deployment. The handover playbook is an example you can open and print.
Next step: choose one process, name its owner and examine a real example together.
welcometotheaijungle.com/method
Questions about AI consulting and managed agents.
Will an agent ever send something to a client without me?
External messages, quotes and publications need the approval defined for the engagement. The reviewer can be a named member of your team: continuity should not depend on the founder approving everything. We define the permissions, exceptions and stop conditions before putting the process into service.
We tried AI tools. They didn't stick. Why would this?
We first examine why the previous attempt failed: unclear ownership, missing context, an unsuitable tool or too much change at once. We choose one process, test it with authorised examples and name the person who will run it. Documentation, a simpler tool or no new AI may be the right answer.
Why not just take the platform?
Choose the platform if your team can configure, review and operate it. Choose a managed partnership when you need the agreed build, continuing maintenance and training. Scope, responsibilities, acceptance criteria and commercial terms are written down. New workflows follow a demonstrated need, not a quota.
How does the revenue share work? Do you look at our books?
The share applies only above the starting point measured at the Assessment. Share measured monthly when the firm closes its books each month: each month is compared with the same month of the year before signing, and a month at or below that level carries no share. Otherwise it is measured once a year, on the contract year. Fixed fee and share together never exceed $15,000 in a month, 3 times the fixed fee. It is based on the firm’s overall growth, not only revenue attributed to AI. The measurement and supporting figures must be agreed and verifiable by both parties. The published rates and signing options are on the pricing page.
What if the AI workflow does not work?
The assessment delivers a reasoned recommendation, alternatives and an action plan reviewed by a human consultant. Before Build, we agree acceptance tests and stop conditions. If a test fails, we identify what needs correcting or keep the existing process. Handover, rights, licences and exit commitments follow the written agreement; ownership of every component is not automatic.
Who is behind AI Jungle?
AI Jungle combines human consulting with AI-supported implementation. A human consultant reviews the paid assessment; the engagement names the people responsible for delivery and approval. You can inspect the Asia-Connect case, our method and the sample handover playbook before discussing a project.