M3 · THE COMPANY BRAIN · P07

The Company Brain Playbook

The OS your dashboard cannot see. Dashboards show outcomes. The brain shows decisions.

11 SLIDES · ~10 MIN · PLAYBOOK OVERVIEW
Key Takeaways
  • Dashboards show outcomes. The brain shows decisions.
  • <10% — of enterprises say their data is fully ready for AI — a consistent finding across HBR Analytic Services and Thales reporting.
  • 34% — know where all their data resides.
  • 40× — open-source AI cost reduction in 18 months. Proprietary data is the only durable advantage.
  • Map the five decisions that recur in your weekly leadership meeting — write them down.
MODULE 3 · PLAYBOOK 07

The Company Brain Playbook

The OS your dashboard cannot see. Dashboards show outcomes. The brain shows decisions.

BuildClub Academy
01 · THE HOOK

The data is the moat — and almost no one is ready to defend it.

<10%
of enterprises say their data is fully ready for AI — a consistent finding across HBR Analytic Services and Thales reporting.
HBR AS / Cloudera 2026
34%
know where all their data resides.
Thales 2026 Data Threat Report
40×
open-source AI cost reduction in 18 months. Proprietary data is the only durable advantage.
THE THESIS

Dashboards show outcomes.
The brain shows decisions.
Build the brain before a competitor indexes yours.

The Thesis
02 · DASHBOARD VS BRAIN

Your dashboard sits on top. The brain underneath is what makes it true.

Dashboard — what
Revenue · Churn · MQLs · NPS · Pipeline. Retrospective reads of the outcome.
Brain — why
Why · Who decided · Tradeoff · What we expected · What we ruled out.
Dashboard alone
Data without institutional memory. You can read it. You cannot train on it.
Dashboard + Brain
Queryable record of the reasoning. The training set for every future agent.
03 · THE 5-STEP PLAYBOOK

Five moves that turn a dashboard company into a brain company.

1
2
3
4
5
1
Map the decisions
Inventory the recurring decisions your top five leaders actually make.
2
Capture the why
Log the reasoning, the tradeoff, and the counterfactual — not just the outcome.
3
Build the retrieval layer
A queryable memory/vector store so agents pull context, not just data. Multiple categories of tool exist (graph memory, vector DB, hybrid); pick one and commit.
4
Wire it to meetings
AI meeting transcripts routed through topic detection into the brain. Most meeting-transcript tools expose an API; the choice is operational, not strategic.
5
Test the query
"Why did we fire that vendor in 2023?" returns a usable answer in 10 seconds.
04 · THE ROUTER PATTERN

How the brain actually routes. One server. Three stores. One answer.

INGESTION automated pipelines keep each store current Meetings · Transcripts → ZEP-SYNC → Zep Graph PDFs · Documents → LlamaParse → Voyage Embed → Supabase (n8n) AI Sessions → Mem0 Nightly Flush → Mem0 Store AI CLIENTS CLIENT Perplexity Computer Browser · agent runtime MCP call → CLIENT Claude Code CLI · dev workflows MCP call → CLIENT Claude Desktop Desktop · daily use MCP call → The MCP server decides which store to query — or queries multiple in parallel MCP SERVER ▲ Vercel EXPOSED TOOLS ▸ search_graph() graph lane ▸ search_documents() docs lane ▸ search_memory() memory lane ◆ ROUTER · DECISION ◆ ? — graph — — docs — — memory — "Who said what?" "What changed since last month?" "Relationship X and Y?" GRAPH MEMORY Zep Cloud Zep Cloud Knowledge graph store → graph.search() Entity + edge query RETURNS ← Entities · Facts · Relationships · Time Knows what changed and WHEN · Bi-temporal "What does this document say?" "Find exact text about Y" VECTOR RAG Supabase pgvector Voyage Embed input_type="query" → pgvector 20 cand · cosine → RETURNS ← Exact text · Page # · Source doc · Top 5 rerank Voyage called twice: embed + rerank "How does our pipeline work?" "What's the config for X?" OPERATIONAL MEMORY Mem0 Mem0 API Portable memory layer → search_memories() Atomic fact lookup RETURNS ← Atomic facts · Pipeline configs · Decisions Portable · Any AI reads with one API call 3 lanes return → ↩ Results converge at MCP Server · Synthesized context ↪ → AI Client generates final answer

Every AI client calls the same MCP server. The router decides which store answers the question — graph, vector, memory — or queries multiple in parallel. Results converge before they return.

05 · WHAT GOES IN THE BRAIN

Six inputs every company already generates — and almost none captures.

1
Decision logs
CEO, CFO, COO recurring calls. Recorded with reasoning, not just outcome.
2
Meeting transcripts
Fireflies, Otter, Zoom AI. The running record of who said what.
3
Customer conversation threads
CRM activity + support tickets + call recordings. The voice of the customer, unfiltered.
4
Competitive intelligence
Analyst notes, win/loss data, pricing intel — as structured records, not slide decks.
5
Vendor evaluations
The reasoning behind platform choices is a future-agent training input.
6
Incident logs
What broke, why, who fixed it, what was learned. The institutional scar tissue.
06 · WHY DASHBOARDS ARE NOT ENOUGH

Everyone has dashboards. Almost no one has a queryable decision history.

Takeaways
  • 1 100% have dashboards. ~60% have data warehouses. ~25% have documented SOPs.
  • 2 ~7% have data completely ready for AI (Cloudera + HBR, Mar 2026).
  • 3 Under 3% have queryable decision histories. The bottom layer is where the moat is — and it is nearly empty.
07 · WHO ALREADY DOES THIS

Three companies that built the brain before anyone called it that.

1
Bridgewater — Principles
Ray Dalio, founder of Bridgewater Associates (the world's largest hedge fund), turned every issue, mistake, and judgment call into a logged entry. The log is the unfair advantage; their AI runs on the principle database.
2
Amazon — PR/FAQ
Every product starts as a press release plus FAQ. Five years of PR/FAQs is a queryable product brain.
3
Boeing — cautionary
Senior engineers retired without structured knowledge transfer; Senate Commerce Committee oversight tied a significant share of 737 MAX certification-related rework and remediation to that lost institutional memory. The headline losses are public record — in the tens of billions across direct costs, settlements, and order cancellations.
08 · THE COMPETITOR INDEX THREAT

Four ways your brain is leaking — to vendors who will index it before you do.

1
Customer data being indexed
The CRM vendor’s AI trains on how you sell. The next buyer of that CRM gets a model partially reflecting your playbook.
2
Support tickets are training data
Every support interaction is pattern data for the vendor’s next product.
3
Contracts are a pricing signal
Every signed contract on a DocuSign-adjacent AI platform is teaching a pricing model.
4
Meetings are being transcribed
Zoom AI, Fireflies, Otter. Who owns the transcript when the vendor’s terms update?

Audit the training-data clauses in your top five vendor contracts this quarter.

MONDAY-MORNING ACTIONS

Five moves to start building the brain this week.

Monday Morning
  1. 1
    Map the five decisions that recur in your weekly leadership meeting — write them down.
  2. 2
    Start recording the "why" on the next three decisions — what you decided and what you ruled out.
  3. 3
    Audit the training-data clauses in your top five vendor contracts — know what you are giving away.
  4. 4
    Stand up a meeting-transcript capture system with defined retention and storage.
  5. 5
    Name a "brain owner" — one person whose job is to make institutional memory queryable.
P07 · Questions CEOs Ask

Frequently Asked Questions

What is the core idea of The Company Brain Playbook?
Dashboards show outcomes. The brain shows decisions.
What does the data say a CEO should pay attention to?
<10% of enterprises say their data is fully ready for AI — a consistent finding across HBR Analytic Services and Thales reporting. 34% know where all their data resides.
What should a CEO do Monday morning after reading The Company Brain Playbook?
Start here: Map the five decisions that recur in your weekly leadership meeting — write them down; Start recording the "why" on the next three decisions — what you decided and what you ruled out; Audit the training-data clauses in your top five vendor contracts — know what you are giving away.
What are the steps in The Company Brain Playbook?
1) Map the decisions; 2) Capture the why; 3) Build the retrieval layer; 4) Wire it to meetings; 5) Test the query.
Where do the claims in The Company Brain Playbook come from?
The playbook cites YPO Technology Network AI Daily Brief — Company Brain (May 16, 2026); HBR Analytic Services / Cloudera AI data readiness research 2026; Thales 2026 Data Threat Report; BuildClub Reference — Enterprise AI Memory Architecture (May 2026).
P07 · Sources

All Sources in This Playbook

00 / 11