Claude OS · Autopilot protocol · 15 role packs

Autopilot. 15 role packs.
The Claude OS distilled from six live businesses.

Brainfile is the Claude Code operating system we run our own six businesses on, packaged for yours. Drop in your role pack, tell Claude what you want and how long it has (the Autopilot protocol), and every session starts with persistent memory, your operating laws, and rules that compound instead of resetting.

Runs on your existing Claude plan. No API costs. The file is the product.

Start your 30-day free trial → See the 7 layers
✓
30-day free trial
✓
30-day money-back guarantee
✓
Cancel anytime
What's inside

7 layers of OS. Click any layer.

The 7 layers below are the stack EP runs to operate six businesses solo — the operating model your Brainfile grows into. Your packs seed the config layers on day one; the rest graduates into customer packs as it proves out in our production. Click each layer to see what's in it — and why it matters.

L1
Core OS
CLAUDE.md · brain/ · memory · command center · session continuity
10 components
▾
What it is: The foundation Claude reads at every session start. Tells Claude who you are, how you work, what's locked, what NOT to ask. Persistent across sessions, version-controlled, self-improving.
CLAUDE.md — Project laws + rules read at every Claude Code start.
brain/ directory — Operational state, KPIs, configs, queues.
Memory index — Persistent cross-session context, never forgets.
Command Center dashboard — One-pane-of-glass for everything.
Session startup — Auto-reads context every new session.
Session handoff — Outgoing session writes notes for the next.
Conversation history — Every message logged + searchable.
Locked decisions tracker — Stop re-answering the same questions.
Priority queue — A/B/C approval buttons for any open decision.
Daily log — Auto-appended record of what got done.
L2
AI Executive Team
Autonomous agents that act as CMO, CTO, CFO, VP CS, VP Revenue
10 agents
▾
What it is: Role-tuned Claude agents defined as plain markdown files. Each one specializes, holds its own context, and runs inside your own Claude sessions on a cadence you set. They coordinate through the brain/ directory.
CMO Agent — Drafts content, runs A/B tests, optimizes funnel.
CTO Agent — Code reviews, refactors, ships infra changes.
CFO Agent — Tracks revenue, costs, cash runway, P&L.
VP Customer Success — Triages support, drafts replies.
VP Revenue — Hunts revenue blockers, fixes broken funnels.
VP Engineering — Build queue execution, PR review.
Strategy Agent — Weekly retros + proactive recommendations.
Research Agent — Always-on competitive + market intelligence.
Backlog Curator — Tracks every promise, never lets you forget.
Inter-agent messaging — Agents coordinate through shared state.
L3
Industry Verticals
Pre-tuned operating systems for your profession
8 verticals
▾
What it is: Each vertical is the full 7-layer OS pre-configured for one profession — its frameworks, vocabulary, decision rules, and workflows already wired in. Pick the one closest to your role and customize.
Founder OS — Fundraising, hiring, pricing, board comms.
Agency OS — Client SOWs, deliverables, project ops.
Creator OS — Content calendar, audience growth, monetization.
Developer OS — Code review, ADR, deploy, on-call.
Operator OS — KPI tracking, retros, cross-functional ops.
Marketer OS — Funnel, copy, A/B test, attribution.
Real Estate OS — Deal analysis, comps, 1031, property mgmt.
Custom vertical builder — Build your own from the OS.
L4
Quality Gates
Code-enforced checks. Bad output never ships.
10 gates
▾
What it is: A wall of automated validators that block low-quality output before it reaches you or your customers. Every "must" in your CLAUDE.md becomes code, not a wish.
Visual QA — Renders pages, catches empty tables + stale data.
Deploy gate — Hard block on bad deploys, fail-closed.
Drift detector — Flags when reality drifts from spec.
Honesty check — Catches "claimed done but isn't" lies.
Spec validator — Output checked against contract.
Brand consistency — Pricing + positioning never drift.
PII scrubber — Personal data never leaks customer-facing.
Content quality scorer — Email + social must score ≥8 to send.
Regression checks — Old features still work after new fixes.
Verify-before-complete — "Done" requires gate-pass, not text.
L5
Self-Improvement Engine
Every error becomes a permanent automated check.
8 components
▾
What it is: The OS gets smarter every day. When something goes wrong, the engine writes a new rule, wires a new gate, and prevents that class of error forever. Compounds over weeks.
Learning engine — Captures patterns from every session.
Error pattern tracker — Spots recurring issues + proposes fixes.
Rule promotion — Validated patterns auto-promote to enforcement.
Session scorecard — Every session graded; trend tracked.
Loop metrics — Throughput, accuracy, cost per task.
Recursive quality — Quality checks audit themselves.
Had-to-ask detector — Spots gaps where you had to repeat.
Weekly retrospective — Auto-generated; ships strategic recs.
L6
Cloud Routines
Scheduled agents that run while you sleep.
12 routines
▾
What it is: Cron-scheduled Claude agents running in the cloud (or locally via LaunchAgents). They process overnight, refresh dashboards, draft content, and queue decisions for your morning review.
Morning brief — Daily summary + priority queue.
Priority scanner — Hourly check on aging tasks.
Marketing daily — Drafts social + email content.
Content research — External market + competitor scan.
Conversation learnings — Extracts feedback patterns.
Feature proposer — Suggests product improvements.
Newsletter generator — Drafts tactical issue daily.
Revenue sentinel — Alerts on revenue anomalies.
Overnight loop — Multi-hour autonomous work block.
Routine meta-tuner — Audits routines for redundancy.
Dashboard rebuild — Nightly fresh data + visuals.
Weekly retrospective — Sunday strategic review.
L7
Integrations
Plugged into your stack out-of-the-box.
12 integrations
▾
What it is: Pre-wired connectors for the tools you already use. MCP servers + API clients + webhooks. No glue code, no scratch builds — connect, authenticate, run.
Stripe — Billing, subscriptions, revenue tracking.
Brevo — Transactional + marketing email at scale.
Buffer — Social posting across X + LinkedIn + more.
Cloudflare — DNS, Pages, Workers, analytics.
Gmail / Google Workspace — Email read + send + alias.
Google Calendar — Booking, scheduling, reminders.
GitHub — PRs, issues, CI/CD via gh CLI.
Anthropic API — Direct API access for batch + custom flows.
MCP servers — Playwright, filesystem, search, DB.
Substack / Beehiiv — Newsletter publishing.
Webhooks — Inbound triggers for any external event.
Custom integrations — Add your own MCP server or API.
Use cases

One OS. Every profession.

Brainfile ships pre-tuned for these roles out of the box. Each is a section of the OS — not a separate product. Switch between them, combine them, customize them.

🚀

Founder & Operator

Fundraising, pricing, hiring, board comms, KPI tracking, weekly retros. The OS runs your business while you sleep.

📈

Finance & Trading

Market regime awareness, risk frameworks, position sizing, earnings analysis, portfolio construction.

🏠

Real Estate Investor

Cap rate, cash-on-cash, DSCR analysis, comps, 1031 exchange logic, property mgmt SOPs.

⚖

Legal & Compliance

Contract review, regulatory awareness (SEC, HIPAA, GDPR), risk matrices, citation standards.

🏥

Healthcare & Clinical

Clinical documentation, evidence-based frameworks, HIPAA-aware comms, research methodology.

📣

Marketing & Growth

Funnel optimization, copywriting frameworks, SEO strategy, email sequences, ad creative briefs.

💻

Software & Engineering

Code review, ADR configurations, deploy + on-call, refactor planning, technical writing.

🎨

Creator & Content

Content calendar, audience growth, monetization, brand voice, newsletter ops.

🤯

Your role

Don't see your profession? Brainfile customizes to any role during onboarding. Same OS, your vocabulary.

Single tier · everything included

Start your 30-day free trial

Full access to all 7 layers, 70+ components, every vertical, every agent. No setup fees. Cancel anytime.

$99 /month

or $999/year — save 16%

Start free trial →

30-day free trial · 30-day money-back guarantee · Cancel anytime

Free starter tool

Try a Brainfile starter — free

Answer 10 questions and we'll generate a starter context file you can drop into Claude today. No charge for 30 days. Want the full 7-layer OS? Start the 30-day free trial above.

Step 1 of 10

What's your role?

This helps Claude understand how to frame answers for you.

Step 2 of 10

Which AI tools do you use?

Select all that apply. Your Brainfile works across all of them.

Claude
ChatGPT
Gemini
Cursor
Copilot
Perplexity
v0/Bolt
Claude Code
Windsurf
Step 3 of 10

How do you like AI to communicate?

Pick the styles that match your preference.

Concise
Detailed
Technical
Plain language
Challenge me
Supportive
Structured
Casual
Step 4 of 10

What do you primarily use AI for?

Select your top use cases.

Writing
Coding
Research
Strategy
Data analysis
Email
Learning
Content
Project mgmt
Design
Step 5 of 10

What are your key technical skills?

Languages, frameworks, platforms you work with.

Step 6 of 10

What are you working on right now?

Active projects, goals, or focus areas this quarter.

Step 7 of 10

How do you make decisions?

This helps AI give you advice that fits your thinking style.

Data-driven
Intuition-first
Consensus
Move fast
Deliberate
Frameworks
Step 8 of 10

What annoys you about AI responses?

Your Brainfile prevents these from happening.

Too verbose
Hedging
Over-cautious
Echoing
Buzzwords
Not asking
Assuming beginner
No reasoning
Step 9 of 10

How should AI format its output?

Defaults that apply to most of your interactions.

Markdown
Code examples
Recommendation first
Tables
Short responses
Include sources
Action items
No emojis
Step 10 of 10

Anything else AI should know about you?

Timezone, name, working hours, unique constraints, context that matters.

We'll never spam you. Unsubscribe anytime. Your starter Brainfile generates either way.

Your starter Brainfile

How to use your starter Brainfile:
Claude: Create a Project → paste into Custom Instructions
ChatGPT: Settings → Personalization → Custom Instructions
Cursor: Save as .cursorrules in your project root
Gemini: Paste at the start of any conversation
Any AI: Paste into the system prompt or first message

Want the full Claude OS?

This starter is one component of Layer 1 (Core OS). The full Brainfile gives you all 7 layers + 70+ components — agents, gates, cloud routines, integrations, and self-improvement.

Start 30-day free trial — $99/mo