Running a Company on EOS — and Why It Made AI Actually Work

Feb 27, 2026

When I joined Ascend, it was an incredibly successful company. $20m ARR, 650+ clients, virtually no churn, and a product that people genuinely loved. But the company completely lacked operational and organisational structure. I ran a Listen & Learn tour — structured conversations with nearly every senior leader over two weeks — and the same four gaps surfaced repeatedly:

Organisation
Nobody knew who owned which project or where accountability fell when things went wrong
Metrics
No way to tell if the business was healthy beyond top-line revenue — no early warning system
Documentation
Decisions scattered across Slack and meetings, so the same discussions happened repeatedly
Priorities
Teams working on different things because there was no shared list of what mattered most

Why EOS

I chose to solve this with a framework called EOS — the Entrepreneurial Operating System. Other frameworks like OKRs focus primarily on goal-setting: define an objective, attach measurable key results, review quarterly. That's useful, but our problems went beyond goals. We needed a single system that addressed organisation, metrics, documentation, and priorities together. EOS covers all four in one interconnected framework.

Each component maps directly to one of the problems above:

Accountability Chart
Organisation
Every seat defined with specific accountabilities, measurables, and a single named owner
Scorecard
Metrics
10–15 weekly numbers with owners and targets — problems surface in days, not months
V/TO + Decision Logs
Documentation
Company identity on one page, every L10 decision logged with context and reasoning
Rocks
Priorities
3–7 quarterly goals, each with an owner and binary done/not done — everything else is secondary

The full system cascades from permanent company identity down to weekly decisions, each layer getting more specific and more frequently updated:

Core Values & Focus
Who we are, what we do, what we don’t
Permanent
3-Year Picture
Where we’re heading
Annual
1-Year Plan
This year’s targets
Annual
Rocks
Top priorities this quarter
90 days
Scorecard
Key metrics & targets
Weekly
L10 + Issues
Decisions, actions, blockers
Weekly

Why this is the perfect context layer for AI

Most companies treat AI as something you bolt onto individual tasks — summarise this document, draft this email, generate this report. That works for isolated tasks. But the moment you want AI to do anything that requires understanding your company — what you're trying to achieve, what was decided last week, what metrics are off track — it needs context.

The bottleneck isn't AI capability. It's that the context doesn't exist in any structured, consistent form. EOS solves this as a side effect of running the company well. Every document it produces — the V/TO, the scorecard, the decision log — is structured by design and updated on a predictable cadence. This is exactly what AI tools need to operate effectively.

EOS documentation
V/TO
Company identity, ICP, brand voice
Rocks
What matters this quarter
Scorecard
What "on track" means, with numbers
Decisions
What was agreed and why
powers
AI tools
Claude Code
Sessions start with full context
Automations
Thresholds from scorecard targets
Alerts
Off-track metrics flagged automatically
Reports
Growth and scorecard analysis

In practice, every Claude Code session starts with the V/TO for ICP definitions and brand voice, current Rocks for quarterly priorities, and scorecard targets for what "on track" means. Automations reference the scorecard directly — an alert fires when a metric misses its target, no human interpretation needed. When the L10 resolves an issue, that decision is logged in Notion with reasoning, and any AI session touching that area references it directly.

The conventional thinking is that structured processes slow you down. With AI, the opposite is true. Structure is what makes speed possible, because it gives every tool the context to operate without constant hand-holding.

The weekly loop

The L10 meeting is where EOS and AI close the loop most concretely. Here's how one actual week works:

Wednesday: A Make.com automation sends the L10 Notion document to Slack. Each leader fills in their wins, headlines, and issues before the meeting — so we don't spend meeting time on status updates.

Monday: The L10 runs. 90 minutes, fixed structure: wins, scorecard review, Rock check-in, headlines, to-do review, then 60 minutes solving the biggest issues. Fireflies transcribes the entire meeting live.

Post-meeting: The Fireflies transcript is processed. Decisions are logged to the Notion decision log automatically. Action items become Asana tasks. Nothing falls through the cracks.

Following Monday: The next L10 reviews last week's to-dos, checks the scorecard, and assesses Rock progress. The loop closes.

Prepare
Wed
Team fills in Notion via Slack
Meet
Mon
L10 runs, Fireflies transcribes
Capture
Auto
Decisions → Notion, tasks → Asana
Review
Mon
Next L10 checks last week’s to-dos
Repeats weekly

Before this, someone had to write up meeting notes by hand, chase people for follow-ups, and hope that decisions made in conversation actually got executed. Now the operating rhythm generates structured data as a byproduct — and that data feeds directly into every AI tool we use.

Record
Capture meetings
Fireflies
Document
Store & organise
Notion
Execute
Track & deliver
Asana
Communicate
Notify & align
Slack
Connected via Make.com automations

Where we are now

Two quarters in. L10s have run every week since October 2025. Rocks evolved from Q4 (implement EOS, identify growth channels, secure partnerships) to Q1 (fix the search bottleneck with Guru, slash concierge overhead by 50%, scale new growth channels to $100k gross profit). The scorecard tracks revenue, growth metrics, concierge SLAs, sprint completion, and runway — reviewed every Monday.

Every AI tool at Ascend plugs into the same context layer. Growth automations reference the ICP and brand guidelines from the V/TO. Guru references client preferences documented through the same system. CRM workflows use scorecard targets to define thresholds. The context compounds — every quarter of structured documentation makes every AI session more effective.

If you're considering EOS, the standard advice is about accountability and focus. That's real. But if you're also building with AI, the structured documentation it produces might be the most valuable part of the whole system.

EOS as a Context Layer for AI.md

EOS as a Context Layer for AI

eos-for-ai.md

Drop this file into your .claude/skills/ directory

/scorecard

scorecard.md

Drop this file into your .claude/commands/ directory

/growth-report

growth-report.md

Drop this file into your .claude/commands/ directory

/ad-review

ad-review.md

Drop this file into your .claude/commands/ directory