--- name: eos-for-ai description: Use when implementing EOS (Entrepreneurial Operating System) as a structured context layer for AI — covering V/TO, Rocks, Scorecard, L10 meetings, and the automation stack that makes it work. --- # EOS as a Context Layer for AI A methodology for implementing EOS so that every document it produces doubles as structured context for AI tools. The framework runs the company well for humans, and the documentation it generates makes AI sessions dramatically more effective. ## When to Use - Implementing EOS at a company for the first time - Already running EOS but want to make it AI-ready - Building AI automations that need company context (goals, metrics, decisions) - Setting up the tooling stack (Notion, Asana, Slack, Fireflies) to capture and route EOS outputs --- ## The Problem EOS Solves Most companies lack structured documentation of four things AI needs to operate: | Gap | Symptom | What AI needs | |-----|---------|---------------| | Organisation | Unclear roles, overlapping ownership | Who owns what, who to ask | | Metrics | No shared scorecard, inconsistent tracking | What "on track" means, with specific numbers | | Documentation | Decisions in Slack threads, nothing captured | What was decided, when, and why | | Priorities | Everything is urgent, no shared focus | What matters this quarter, in priority order | EOS addresses all four in a single interconnected system. --- ## Step 1: Set Up the V/TO (Vision/Traction Organiser) The V/TO is a single document that captures the permanent and long-term layers of the company. It becomes the living brief that AI references for company context. ### Fields to complete ``` Core Values (3-5): 1. ___ 2. ___ 3. ___ Core Focus: Purpose / Cause / Passion: ___ Niche: ___ 10-Year Target: ___ Marketing Strategy: Target Market: ___ Uniques (3): ___ Proven Process: ___ Guarantee: ___ 3-Year Picture: Revenue: ___ Profit: ___ What does it look like? ___ 1-Year Plan: Revenue goal: ___ Profit goal: ___ Measurables: ___ ``` ### AI integration Store the V/TO in Notion (or equivalent). Reference it in every Claude Code session that touches strategy, messaging, growth, or product. The V/TO answers: who are we, who do we serve, what do we stand for, and where are we going. --- ## Step 2: Define Rocks (90-Day Priorities) Rocks are the 3-7 most important goals for the next quarter. Each has an owner and a binary definition of done. ### Template per Rock ``` Rock: ___ Owner: ___ Due: End of Q[X] [Year] Definition of done: ___ Status: [On track / Off track / Complete] ``` ### Rules - Maximum 7 company Rocks per quarter - Each person owns no more than 3-5 Rocks - "Done" must be binary — no "80% complete" - Review progress weekly in the L10 - Carry-over is a red flag, not a habit ### AI integration Rocks act as a prioritised backlog for AI sessions. When building features, automations, or reports, reference current Rocks to ensure work aligns with quarterly priorities. --- ## Step 3: Build the Scorecard The scorecard is 10-15 weekly metrics that tell you whether the business is healthy. Each has an owner and a target. ### Template ``` | Metric | Owner | Target | Frequency | |--------|-------|--------|-----------| | Weekly Gross Revenue | [Name] | $___k | Weekly | | Weekly Net Revenue | [Name] | $___k | Weekly | | New Leads (Qualified) | [Name] | ___ | Weekly | | Conversion Rate (MQL → Meeting) | [Name] | ___% | Weekly | | Median Response Time | [Name] | ___s | Weekly | | Customer Satisfaction | [Name] | ___/10 | Weekly | | Sprint Completion % | [Name] | ___% | Weekly | | Runway (months) | [Name] | ___ | Monthly | ``` ### Rules - Review every metric every week in the L10 - Off-track metrics (missing target 2+ weeks) become Issues - Don't add metrics you won't review — keep it to 10-15 - Targets should be stretch but achievable ### AI integration Automations reference scorecard targets as thresholds. An alert that fires when a metric is off track uses the scorecard target directly — no human needs to define what "off track" means. The `/scorecard` command analyses weekly data against these targets automatically. --- ## Step 4: Run the L10 Meeting The L10 is a weekly 90-minute leadership meeting with a fixed agenda. It generates the most valuable AI context: decisions, action items, and issue resolutions. ### Agenda (fixed) ``` 5 min — Segue (good news, personal/professional) 5 min — Scorecard review 5 min — Rock review 5 min — Customer/employee headlines 5 min — To-do review (from last week) 60 min — IDS (Identify, Discuss, Solve) the issues list 5 min — Conclude (recap to-dos, rate the meeting 1-10) ``` ### IDS process For each issue on the list: 1. **Identify** the root cause (not the symptom) 2. **Discuss** (set a timer — prevent rabbit holes) 3. **Solve** with a concrete next step: a to-do with an owner and a deadline ### AI integration — the weekly loop ``` Friday: Automation sends L10 template to Slack. Leaders fill in wins, headlines, issues before the meeting. Monday: L10 runs. Fireflies (or equivalent) transcribes live. Post-meeting: Transcript processed automatically. Decisions → Notion decision log. Action items → Asana tasks. Next Monday: L10 reviews last week's to-dos. Loop closes. ``` This loop means every decision and action is captured, routed, and reviewable by AI. --- ## Step 5: Wire Up the Automation Stack The tools that make EOS outputs machine-readable: | Stage | Tool | What it captures | |-------|------|-----------------| | Record | Fireflies (or Otter, Granola) | Full meeting transcript | | Document | Notion | V/TO, scorecard, decisions, issues | | Execute | Asana (or Linear, Jira) | Tasks, Rocks, to-dos | | Communicate | Slack | Notifications, pre-meeting prep | | Connect | Make.com (or Zapier) | Routes data between tools | ### Key automations 1. **Pre-L10 prep**: Make.com sends the Notion L10 template to a Slack channel on Friday 2. **Post-L10 capture**: Fireflies transcript → parsed for decisions and action items → Notion + Asana 3. **Scorecard alerts**: When a metric misses target for 2+ consecutive weeks, Slack alert to the owner 4. **Rock status**: Weekly Asana report on Rock completion status --- ## Step 6: Make It AI-Ready Once EOS is running, every AI session can reference: - **V/TO** for company context (ICP, brand voice, positioning, strategy) - **Rocks** for current priorities (what matters this quarter) - **Scorecard** for targets and thresholds (what "on track" means) - **Decision log** for recent decisions (what was agreed and why) - **Issues list** for known problems (what's being worked on) ### Claude Code session template At the start of any work session, load: ``` Context: - V/TO: [link or paste] - Current Rocks: [list with owners] - Scorecard targets: [key metrics and numbers] - Recent decisions: [last 2-3 L10 decisions relevant to this work] ``` This eliminates the cold-start problem. The AI doesn't need you to re-explain the company — EOS has already documented it in a structured, consistent format. --- ## Checklist - [ ] V/TO completed and stored in Notion - [ ] 3-7 Rocks defined with owners and definitions of done - [ ] Scorecard built with 10-15 metrics, owners, and targets - [ ] L10 meeting scheduled weekly (same time, same day) - [ ] Fireflies (or equivalent) recording meetings - [ ] Make.com automations routing decisions and tasks - [ ] AI sessions reference V/TO, Rocks, and scorecard - [ ] `/scorecard` command set up for weekly analysis