Business Brain Prompt: Give AI a Shared Company Brief

A business brain prompt is a structured brief you give an AI assistant at the start of a session so it understands who the company is, what it is currently working on, and which rules govern how it should behave. Without your brief, every new session starts from zero — the AI knows nothing about your industry, your terminology, your current priorities, or the decisions people have already made. Tools like fenbs are designed to keep that brief current and accessible to every connected assistant, so the work continues from the same foundation rather than requiring a new explanation each time.

Your brief covers five components: company goals, working definitions, rules and constraints, current projects and priorities, and open decisions. You do not need all five in every session — choose the components that affect the session's task. But having all five written down and maintained means you can always pull the right section rather than starting from memory.

Why a company brief improves AI output

Without a brief, an AI assistant working on a company task makes assumptions: what the company sells, who its customers are, what terms mean internally, what has already been decided. Some of those assumptions will be wrong. A brief replaces assumptions with facts the team has agreed on, which produces output that is relevant to your actual situation rather than to an imagined average company.

A brief also prevents repeated correction. If you find yourself telling an AI "that's not quite right for our business" in the third message of every session, the information you are providing in that correction belongs in your brief. Move it there once and stop repeating it.

The most important part of a good brief is the distinction between facts and decisions. A fact about the company — "we serve small manufacturing businesses in the Midwest" — belongs in your brief as context. A decision — "we will not pursue customers outside the US this year" — belongs there as a rule, with the person who decided it and the reason. The AI treats these differently: context is background, a rule is a constraint it must follow.

The five components of a business brain prompt

The full company brief covers five areas. Each one serves a different purpose in the session.

• Company goals: what the business is trying to achieve this quarter or year, in plain terms. Not a mission statement — a current priority.
• Working definitions: terms the company uses in a specific way that differ from their common meaning, or internal shorthand the AI would otherwise misread.
• Rules and constraints: decisions the team has made that limit how problems can be approached. These are standing rules, not preferences.
• Current projects and priorities: what is actively being worked on and what the AI is being asked to help with today.
• Open decisions: questions that have not yet been settled, so the AI knows not to treat them as resolved.

Here is the company brief for this session. Read it before responding to any task I give you.

COMPANY GOALS (current quarter):
[List 2–4 specific goals, e.g., "Increase trial-to-paid conversion rate" or "Launch three new customer segments by Q4"]

WORKING DEFINITIONS:
[Define terms used internally in a non-standard way, e.g., "'A lead' in our usage means a company that has booked a demo, not just submitted their email"]

RULES AND CONSTRAINTS:
[List standing decisions, e.g., "Do not propose solutions that require a price increase above X%", "All customer-facing content must be approved by Legal before use"]

CURRENT PROJECTS:
[Describe what is actively in progress and relevant to today's session]

OPEN DECISIONS:
[List questions not yet settled, e.g., "We have not decided whether to expand to the Canadian market this year — treat this as open"]

For this session, your task is: [describe the specific task]

How to write effective rules and constraints

Rules are the most important and the most commonly written poorly. A rule that says "be professional" gives the AI nothing useful — every model thinks it is being professional. A rule that says "do not recommend solutions that require hiring more than two additional staff members before we reach profitability" is specific enough to actually constrain the output.

Write each rule as a constraint, not an aspiration. Include the scope (does this rule apply to all tasks or only to customer-facing work?), the reason if it helps, and the person or decision that established it. The reason is useful when a session produces an edge case where the literal rule and the intended outcome diverge — the reason helps the person reviewing the output decide which way to go.

Here are notes from a team discussion about [topic]: [paste notes]

Extract the decisions that should become standing rules for AI work in this area. For each rule you identify:
1. Write the rule as a constraint in one sentence.
2. Name who made the decision if it is clear from the notes.
3. Add the reason if the notes explain it.
4. Flag anything that sounds like a preference rather than a settled decision.

Do not include anything that is still under discussion.

How to keep your brief current

A brief that is not maintained becomes a liability. If your brief says the company is not pursuing a Canadian market and the company quietly started a Canadian pilot last month, every session that reads the old brief is working from wrong information. A brief that becomes unreliable is worse than no brief at all, because it gives the assistant false confidence.

The practical solution is to update your brief when a decision changes, not at a scheduled review. When someone settles an open decision, close it and move it to the rules section. When a project completes, remove it from current projects. When a rule is superseded, replace it with the new decision and note what changed. Treat your brief as a living document that reflects the current state, not a record of everything that was ever decided.

Prompts for working decisions into your brief

Open decisions in your brief represent uncertainty — places where the team has not yet settled on an answer. An AI should not resolve these itself. The prompt for handling open decisions tells the assistant to surface them when relevant rather than choose between options it was not given authority to choose between.

The company brief lists this as an open decision: [describe the open decision]

In your response to my task, note if this open decision affects the options or recommendations you would make. If it does, show two versions of your response: one if the answer is [option A] and one if the answer is [option B]. Do not choose between them — present both clearly so a person can decide.

When a decision is made, update your brief. Move it from the open decisions section to the rules section with the person who decided, the date, and the reason. The AI then reads the settled answer in every future session rather than presenting both options again. Read system prompt examples and prompt engineering for more on structuring instructions that hold across sessions.

How to do this with fenbs

1. Create a free account and add a project for your company's AI work.
2. Write the full company brief using the template above. Paste the goals and current projects into the project context notes.
3. Add each standing rule as a separate Decision in the project, with the person who decided it, the date, and the reason.
4. List open decisions as context notes marked clearly as open — "not yet decided." Update them when the team settles on an answer.
5. Connect your AI app over MCP from Integrations. Ask the assistant to read the project context and confirm it understands the current goals and rules before starting any task.

fenbs does not write company strategy or make business decisions. It keeps your brief that tells connected AI apps what the team has already decided, so they work from the current state of the business rather than a blank assumption. For a deeper look at what an AI brain for a business can hold beyond your brief, see what is an AI brain.