STRATEGY REVIEW

AI strategy review: pressure-test the memo before the meeting

Bring the board memo, the plan or the deal notes. Up to 10 models from different labs attack it independently, and a moderator writes one verdict with the strongest objections first and the minority view kept.

Find the hard questions before the meeting See plans and limits

AI strategy review on Council AI means giving a board memo, strategy plan, investment note or due-diligence summary to several models from different labs, having each critique it independently, and reading one verdict that ranks the objections and keeps the dissent. One model asked to review a plan tends to be agreeable and to soften its objections; several independent reviewers from different labs disagree with each other, and the disagreement is where the weak assumptions show. On decision questions the models are asked to commit to a recommendation rather than hedge, and the verdict says whether the council converged or stayed split. It is a way to find the hard questions before the meeting, not financial, legal or investment advice.

What a council asks that a friendly reviewer does not

Load-bearing assumptions. The one growth rate, price or hiring plan the whole memo depends on, and what happens to the conclusion if it is wrong.
Missing alternatives. Options the memo never considered, including doing nothing, and why a reader might prefer one.
Risks written as footnotes. Concentration in one customer or supplier, regulatory exposure, key-person risk, a covenant that bites in the downside case.
Internal inconsistency. A budget that does not match the hiring plan, a timeline that contradicts a dependency named two pages earlier.
The question the board will ask. Where the models disagree about the recommendation itself, which is usually the question you will be asked in the room.

A worked example

Ask for the critique you would get from a sceptical board, and for a decision:

This is a board memo recommending we enter the German market next year. Act as a sceptical board member. List the five strongest objections in order of how likely they are to change the decision, the assumption each one attacks, and what evidence would answer it. Then say whether you would approve, delay or reject, and why.

Each model commits to approve, delay or reject. The verdict says how the council split, leads with the objections most models raised, and lists the case for the minority recommendation instead of averaging it away.

How it works

Bring the finished work. Paste it or attach it, and say what you want checked. A clear instruction ("find what is wrong with this") gets a review; a vague one gets a summary.
Several models review it independently. Each model in your council, from a different lab, writes its own review. The first round is blind unless reading each other clearly helps, so no review is a copy of another.
One verdict, with the dissent listed. A moderator model reads every review and writes one verdict. Points most models raised come first; a point only one model raised is listed under it instead of being averaged away.

Limits worth knowing

Council size: up to 7 models per question on Plus, up to 10 on Pro and Ultra, picked from 21 models across 9 labs.
Files per message: Plus 4 files of up to 10MB each, Pro 5 of up to 20MB, Ultra 10 of up to 50MB.
Text per message: about 200,000 characters of extracted text per file and across all files in one message (roughly 120 pages). A file over that is refused with a message saying so, never silently cut.
Agreement is a stronger signal, not a guarantee. Several models can share the same mistake, so the verdict lists the dissent rather than hiding it.
Supported files include PDFs, Word documents, images of slides, and plain text.
On paid plans, Projects keep a deal's chats and standing instructions together, so every review of the same plan starts from the same context.

When a council is the wrong tool

Decisions that need data the models cannot see (your actual numbers, your market, your people), or replacing the judgement of advisers who are accountable for the outcome.

Important

Not financial, legal or investment advice. Models can be wrong, including all of them at once. Use the verdict to sharpen the questions you take to accountable advisers.

Contract review: The term sheet or agreement behind the deal, read clause by clause by several models.
Research verification: Check the market claims and figures the memo relies on.
Plans and limits: Council size, file limits and monthly budget for each plan.

Frequently asked questions

Can AI review a business plan or board memo?

It can critique one: find load-bearing assumptions, missing alternatives and inconsistencies, and argue the other side. A single model tends to be agreeable, so Council has several models from different labs review it independently and shows where they disagree about the recommendation itself.

Will the models just agree with my plan?

Less than one model would. Each reviews independently in the first round, and on decision questions each is asked to commit to a recommendation. When the council splits, the verdict says so and keeps the minority case.

Is this investment or financial advice?

No. It is a critique of the document you provide, by language models that can be wrong. Use it to prepare better questions for advisers who are accountable for the decision.

Can I keep a deal's documents together?

On paid plans, Projects keep related chats and project instructions together. On Ultra, the Library can also search stored documents, retrieving the most relevant excerpts for each question rather than reading every file in full.

Can I try it before paying?

Yes. New accounts get 3 councils free, enough to see three models disagree and read the verdict. Plus ($19.99/mo) is the entry plan with every premium model and councils of up to 7 models; Pro ($39.99/mo) runs 10 models with priority routing. Cancel anytime.