AI Engine/Intelligence/Founder + Leadership Context

Intelligence 07

Founder + Leadership Context

Internal context is not customer evidence.

Leadership knows the company's history, ambition, constraints, and appetite for risk. That context belongs in the intelligence system. Its status must stay visible so a belief does not quietly become a fact, a temporary constraint does not become permanent strategy, and a decision does not get reopened in every meeting.

Its boundary: Intelligence preserves what leaders know, believe, and have decided. Strategy weighs that context against external evidence and chooses the move.

01Intelligence02Strategy03Creation04Distribution05Pipeline06Lifecycle07Operations08Measurement

The Context Ledger

The most dangerous sentence is a belief stored as a fact.

Executive statements carry different kinds of authority. A company goal is a choice. Available cash is a constraint. A claim about buyer behavior is a hypothesis until customer evidence supports it. An approved decision governs current work, but it is not a universal truth.

Classify each important statement when it enters the system. Preserve its owner, source, date, and review rule. AI can then retrieve the right context without flattening everything into one summary.

The Reconstruction Tax

Undocumented context becomes a meeting subscription.

When goals, constraints, and past decisions live inside people's heads, every new campaign begins by rebuilding the same picture. The visible cost is a recurring alignment meeting. The larger cost is execution time lost before and after it.

The simple example below shows staff hours only. It excludes preparation, follow-up, context switching, and payroll.

Decision Half-Life

A decision should expire when its reason does.

Teams usually make one of two mistakes. They relitigate a sound decision whenever someone has a new opinion, or they protect an old decision after its assumptions have changed. A decision record solves both problems.

Record the rationale, owner, scope, effective date, and the events that require review. No trigger means no reopening. A trigger means the original decision deserves a fresh look, not automatic reversal.

The Contradiction Report

AI should find contradictions, not decide who wins them.

A strategy document can say enterprise while the active work remains self-serve. The operating plan can forbid hiring while the campaign plan assumes three specialists. These conflicts often survive because each statement lives in a different tool or meeting.

An AI system can compare approved direction, budgets, decisions, and work in motion. It should surface the exact conflict with citations. A named person still decides whether to cancel the work, revise the direction, or approve an explicit exception.

The Leadership Brief

One page should explain why the current work exists.

The brief is not a company wiki or a meeting transcript. It is the smallest current record a marketing team needs to act without reconstructing leadership intent. Every field has an owner. Every factual statement links to its source. Every unresolved belief stays labeled as a hypothesis.

01

Preserve the source

Meeting summaries can help discovery. Material direction and decisions should link back to the exact transcript, document, or approved record.

02

Keep dissent visible

A summary that erases disagreement also erases risk. Preserve material contradictions until the decision owner resolves them.

03

Assign one owner

Many people can contribute context. One person owns the decision, its review triggers, and the consequences inside the declared scope.

Current Tools

Capture conversation, then promote decisions into a durable record.

Meeting capture and company search reduce retrieval work. Neither replaces the approved source. The best setup separates raw conversation, current decisions, cross-source analysis, and final ownership.

01
NotionDefault source of record
Combines meeting capture, searchable pages, databases, permissions, and cited workspace search. A decision can move from transcript to an owned, reviewable page without leaving the system.
Best fitSeed through Series B teams willing to maintain one shared operating record. Meeting transcription requires participant consent and the right data controls.
02
GranolaFocused conversation capture
Combines typed notes with meeting transcription and AI-assisted summaries. It is useful when leadership conversations contain nuance that a short action list would lose.
Best fitLeaders who want strong personal meeting notes before selected decisions are promoted into the company record.
03
ChatGPT Company KnowledgeCited cross-source analysis
Searches connected company sources and returns organization-specific answers with citations. It can compare direction, decisions, and work across tools when permissions and approved sources are configured correctly.
Best fitTeams on ChatGPT Business, Enterprise, or Edu that already keep reliable records across several connected systems.
04
Confluence DecisionsStructured decision history
Records decision status, impact, participants, options, outcome, and action items in a searchable log.
Best fitCompanies already operating in Atlassian that need clearer decision ownership without adopting a new central workspace.

My default: Notion as the maintained source of record, Granola for high-value conversation capture, and ChatGPT Company Knowledge for cited synthesis across approved systems. If Confluence already runs the company, improve the decision record there instead of creating another repository.

Examples Worth Studying

Three operating ideas that keep context from becoming bureaucracy.

These examples show useful methods. The right implementation depends on company size, risk, and working style.

Public operating handbook

GitLab: DRI + async records

GitLab assigns a directly responsible individual to decisions and directs important decisions and discussions into durable written systems so people can contribute context without requiring consensus.

Lesson: broad input and single-person accountability can coexist when the decision right is explicit.

Read the source
Published operating method

Amazon: Working Backwards

Amazon's published Working Backwards material describes writing a press release and FAQ before development so the proposed customer outcome and hard questions become visible before resources are committed.

Lesson: force leadership intent into an inspectable document before execution makes ambiguity expensive.

Read the source
Published decision framework

Atlassian: DACI decision log

Atlassian's decision template records a driver, approver, contributors, informed parties, status, options, outcome, and follow-up work.

Lesson: a decision is easier to execute and harder to rewrite when the authority and rejected alternatives remain visible.

Read the source

Tool sources: Notion AI Meeting Notes, Notion Enterprise Search, Granola, ChatGPT Company Knowledge, and Confluence Decisions.

Example sources: GitLab decision making, GitLab communication handbook, Amazon Working Backwards, and Atlassian DACI.

Tools, links, and rankings reviewed Q3 2026. Recheck current capabilities, permissions, recording consent, and data policies before procurement.

Intelligence 07

Preserve leadership intent without turning it into market truth.

Classify the statement. Keep its source and owner. Record decisions with review triggers. Then let AI retrieve the context, expose contradictions, and return execution time without deciding what the company should believe.