A virtual board of directors: getting six views on one decision
The head of a small company has no board where an owner looks at the money, an HR director at the people and a lawyer at the risks — so decisions get made from one point of view. Below is the technique that replaces it, and why the value lies in the disagreements.
The head of a small company has no board where an owner looks at the money, an HR director at the people and a lawyer at the risks — so decisions get made from one point of view. Below is the technique that replaces it, and why the value lies in the disagreements.
---
A decision with nobody to discuss it with
The head of a small company, or a freelancer working alone, has a task with nothing to close it: nobody to discuss a decision with.
Not because they know nobody. Because a proper analysis needs people with different points of view — and the ones around them see the situation the way they do, or are not in the field at all.
In a large company that role is played by a board: the owner looks at money, the operations director at manageability, HR at people, the lawyer at risk. Each sees their own, and the decision comes out three-dimensional.
You do not have that. So you make decisions from one point of view — yours — and discover what you missed after the fact.
The technique that replaces it is below. Nobody in the niche writes about it at all.
How it works
The same question gets put to several assistants with different roles.
Not "give me different opinions" to one assistant — that produces a polite list. But a separate run with each role, because each has its own frame, its own criteria and its own questions for you.
The difference is fundamental. A legal assistant will ask about the contract and liability. An analyst about figures and hypotheses. An HR one about people and resistance. None will ask the others' questions.
Six positions worth covering:
The owner — what happens to the money, will it pay back, where is the risk of loss.
The operational view — how manageable is this, who will do it, what breaks.
People — how will the team react, who will sabotage it and why.
Legal — what obligations arise, where liability sits, what goes in the contract.
Market and client — how this looks from outside, what a buyer will say.
Analytical — which figures support this and which contradict it.
The sixth position is particularly useful last: it checks what the first five said.
The order of the runs
To get an analysis rather than six similar texts.
Step 1. State the decision concretely
Not "should I expand" but "should I take on a second contractor for the current volume of work".
A vague question produces vague answers from all six roles — and the analysis turns into general reflection.
Step 2. Give identical inputs
Every role gets the same context. Otherwise you are comparing not positions but different understandings of the situation.
Step 3. Run them one at a time
A separate chat per role. Not in one conversation — otherwise the later answers adjust to the earlier ones.
Step 4. Collect the disagreements
This is where the value appears. Not where they all agree but where the positions contradict each other.
The owner says "it pays back in two months", HR says "there are no people and there will not be". That is the fork you would not have seen alone.
Step 5. Decide yourself
The assistants do not vote and do not make the decision. They show you the field. The choice is yours, and that is fundamental.
What this technique finds
Three kinds of finding that do not appear in a single analysis.
A blind spot by role. You think like a freelancer — and do not think about the legal side. Or like an entrepreneur — and do not think about who will do it with their hands.
Contradictions in your own plan. A plan looks coherent while you look at it from one side. Two different frames quickly find the places where one excludes the other.
Questions you had not asked yourself. Every role starts with its own clarifying questions. Half the value is in those, before any answers.
A related technique: analysing the failure in advance
Worth mentioning separately, because they work as a pair.
Instead of "what are our risks", a different question gets asked: six months have passed, the project failed spectacularly. Why?
The difference is psychological and decisive. A person is no longer speculating — they are explaining something that already happened. That technique uncovers roughly a third more real threats than the direct question.
And a second effect: doubts get legitimised. Naming the cause of a failure that "has already happened" is not criticism but doing the exercise.
How to combine it with the board: first the six positions map the field, then the failure analysis tests the chosen decision for strength.
Where the technique does not work
Honestly, so it does not get overrated.
It does not replace data. Six opinions on empty inputs give you six competent pieces of reasoning about nothing. If you have no figures, get figures first.
It does not remove responsibility. The decision stays yours, and "the assistant said so" will not work as a defence.
It does not suit small decisions. A run takes time. For "which colour should the button be" it is excessive.
And it needs honest inputs. If you describe the situation the way you would like it to be, all six roles will confirm what you had already decided.
How to use this with a client
Here is a non-obvious application, and it makes money.
An analysis of a decision is a service in itself. Not a consultation but a structured document: six positions, the disagreements between them, the forks, what to check before starting.
How to approach it. You come to a client not with an offer of services but with an analysis of their own planned move. Three precise observations about their situation work better than any commercial proposal.
What gets delivered: the positions by role, a map of the disagreements, a list of questions with no answers, and a priority — what to check first.
The price range. By my scrape of 848 listings, a system or pipeline runs $490–1 460 against a $30–75 median for one-off work. Analytical work with a document at the end belongs to the second category.
What is available on the platform
More than 140 AI assistants across 14 categories, and the board's roles come straight from there: entrepreneurs, strategic analytics, legal, HR, sales, marketing.
Strategic analytics — twelve assistants, and most of them are decision-making methods: analysing a failure in advance, devil's advocate, six thinking hats, design thinking, Cynefin, the Pareto method, blue ocean strategy.
That is not a "clever chat" but a run through a method — you get the result of a procedure rather than an opinion.
Refinement modes on most assistants: harsh assessment, focus group, expert review, criteria-based refinement. For analysing decisions the harsh assessment is the most useful — it finds the weak points in your own reasoning.
AI Workflow — if an analysis like this is needed regularly, the six runs get assembled into one chain. Each role's result goes into a shared document.
Chat access comes with the Basic plan. Registration is free and opens three days of full Basic.
Where to start today
One action. Take a decision you are currently putting off — the one where it is unclear what is right.
State it as one concrete sentence and put it to two assistants with different roles: the one that looks at money and the one that looks at people.
Not six. Two.
Usually the disagreement between those two positions is itself the explanation of why the decision is being put off.
---
Where each position's quality comes from
A board of six roles works exactly as well as each role can reason rather than produce the first thing that comes.
Module 4, "Advanced prompting", covers this as sixteen methods in four groups: fundamentals and context · logic and reasoning · multi-step processes · autonomous methods.
The second group is directly about this: how to make a model show its working rather than only its conclusion. In analysing a decision that is critical: you need to see why a role reached that conclusion, or six opinions become six unverifiable assertions.
And a technique from the bot architect — a reasoning chain of seven steps: Understand → Basics → Break Down → Analyze → Build → Edge Cases → Final Answer.
The second-to-last step, edge cases, is usually exactly what a board gets convened for: it finds what nobody thought about in the main scenario.
Module 4 — Basic.
---
What to read next
[A pre-mortem: analysing the failure in advance](/en/blog/ai-for-analyst-premortem) — the paired technique for testing a decision.
[One assistant against three](/en/blog/one-assistant-vs-three) — the mechanics of chains in general.
[AI for business owners](/en/blog/ai-for-business-owner) — where decisions like this get made.
[Six brief fields](/en/blog/six-brief-fields) — what to give each role as input.
[AI agents: what they are and what people pay](/en/blog/ai-agents-guide) — the full guide to agents, with market rates.