LIVING OFF AI · THE FOUNDATION IS BUILT

Maintenance and Growth

Your system now holds context and learns weekly. Two jobs remain, forever: keep it lean, and make it grow. Fifteen minutes every Sunday covers the first. Modules 4 and 5 are the second.

MAINTAIN15 min every Sunday+ 2 audits, monthlyGROWModule 4: tools + skillsModule 5: expert work
MAINTAIN · WHY

From 30 Files to 3,000

Mine grew from ~30 files to 3,123 in three months. You cannot load it all, and you should not.

Growth is not the problem. Silent rot is: stale indexes, duplicate files, loose ends left in chat.

📎 Count: one builder's live vault, Jul 2026, archive excluded.

AprMayJul~30 files3,123 filesa context window fits ~1%indexes decide what loads
MAINTAIN · SAME PROTOCOL, NEW FOLDER

One Index Per Folder

The CONTEXT protocol from last deck, now on projects/: exactly one index, everything listed, one line each. Unlisted = invisible.

Same rule, every folder, forever. A trigger word checks any folder against its index.

📎 The 3-layer canon from last deck, enforced. Index sizes: small folder = 10 lines, my network folder = 2.4k tokens.

📁 projects/_projects-CONTEXT.md · one per folder ✓client-a.md · listed ✓client-b.md · listed ✓notes-tuesday.md · NOT listed → invisible_projects-CONTEXT-old.md · second index → conflictTwo failure modes. Both silent. Both found by the audit.
MAINTAIN · THE ROUTINE

Fifteen Minutes Every Sunday

Four moves, AI-prepared, you judge: review the week · approve the focus update · archive what finished · check the size budgets.

Mine arrives inside the Sunday email from last deck. Three replies from the phone, ~12 minutes.

📎 Weekly-review skill: materials/library · budgets: ~1,000 chars per life area, ~14 KB focus total · overflow → dated logs, never deleted.

15minutes1 · review the week ✓2 · approve focus update ✓3 · archive what finished ✓4 · check size budgets ✓OK 1 3 5 → EDIT 4: … → OKthree replies from the phone · Sunday done
MAINTAIN · DAILY

Close Every Session

Chat is where decisions go to die. Before ending a work session, the AI answers two questions: does any index need updating? does the focus file need updating?

Mine is a trigger word. Clean exit or a numbered list.

📎 My trigger: one word at session end. Cost: ~30 seconds when clean.

session end · exit checkQ1 index update needed? ..... noQ2 focus update needed? ..... YES→ proposed: business area, 1 line "demo day confirmed Jul 28" OK?me: OK → ✅ exit clean
AUDIT 1 OF 2 · THE MACHINERY

The System Audit

Monthly, the AI audits its own machinery: what loads and what it costs · skills used and unused · every tool tested live · security holes · honest comparison with the field.

Read-only. It reports, you decide.

📎 Skill file: library/system-audit.md · prompt on the next slide.

1 · MAP — what loads, in tokens2 · SKILLS — used, unused, missing3 · TOOLS — test each one, live4 · SECURITY — holes, keys, who connects5 · COMPARE — where am I behindoutput: top 5 fixes, ordered by impact
AUDIT 1 OF 2 · PASTE THIS

The System Audit Prompt

Run a full audit of my second brain. Work in parallel agents if you can. 1 MAP — What is this system? Which files load automatically at session start? Measure each in tokens. Lean or bloated, current or stale, and do any two contradict each other? 2 SKILLS — List every skill/trigger. Which were unused for 30+ days? Compared to what people publicly build today, name up to 3 I lack. 3 TOOLS — List every connected tool: MCP servers, APIs, scheduled jobs, channels. TEST each one live. Flag broken jobs, dead keys, silent channels, tools connected but never used. 4 SECURITY — Where are the holes? What can connect from outside? Which credentials sit in plain text? Which tools can write or send externally without my approval? What would you tighten first? 5 COMPARE — Against the current state of the art, where am I behind? Report the five sections, then TOP 5 FIXES ordered by impact. Read-only: change NOTHING in this run.

📎 Run it on the strongest model at maximum thinking, parallel agents on. Mine took ~30 min on ChatGPT and ~50 min on Claude Code. Full skill: library/system-audit.md

AUDIT 2 OF 2 · THE WRITING

The Content Audit

The librarian pass: every folder has one honest index · constitution measured against its own token budget · all 5 layers present, nothing leaked between them · focus file matches reality.

Output: merge, archive, rewrite. Proposed only.

📎 Skill file: library/content-audit.md · prompt on the next slide.

walk every folder …📁 projects/ · index ✓ · all files listed ✓📁 health/ · NO index · 9 files invisible📄 constitution · 11.2k tok / budget 12k ✓ · 1 stale date🗂 5 layers · present ✓ · 3 skills leaked into constitution→ MERGE 1 · ARCHIVE 2 · REWRITE 2changes proposed, nothing touched
AUDIT 2 OF 2 · PASTE THIS

The Content Audit Prompt

Audit the CONTENT of my second brain, not the machinery. 1 INDEXES — Walk every folder holding more than a handful of files. Each must have exactly one index file. Flag folders with none, folders with two, and orphan files listed in no index. 2 CONSTITUTION — Measure every always-loaded file in tokens. Compare against the budget written in the constitution itself. Flag stale dates, ended projects, statements a newer file contradicts. 3 FIVE LAYERS — Constitution, memory, skills, tools, focus: each present, findable, current? Flag content leaked into a wrong layer. 4 FOCUS — Does the focus file match reality? Dead dates, finished projects still listed, new work it never mentions? Report per folder, one line each: OK or the finding. End with one list: MERGE / ARCHIVE / REWRITE, one line per item. Propose only. Change NOTHING yet.

📎 Same rule: strongest model, maximum thinking, 30 to 50 minutes. Full skill: library/content-audit.md

GROW · WHAT COMES NEXT

The Growth Ladder

Four rungs. Each one stands on everything below it, and each adds cost that must pay for itself in time or revenue.

The next five slides show the top rung in practice: machines built on a second brain that people pay for.

📎 The business case: the money deck (m1-4) · rungs map to the course modules.

MODULES 0–3 · FOUNDATION · done today5 layers · the loop · Sunday routine · 2 auditsMODULE 4 · TOOLS + SKILLSthe vetted MCP list · calls & email analysis · skills you writeMODULE 5 · APPS + AUTOMATIONSapp builders like Lovable · agent teams · costs grow with valueEXPERT · SELLABLE MACHINESnext 5 slides · work clients pay for
GROW · EXAMPLE 1 OF 5

A Client Website Machine

Client calls live in the brain, analyzed. From them: a one-page brief in the client’s own words. An app builder ships the site from that brief.

I shipped my agency’s site this way, bilingual, in days.

📎 Pipeline: call analysis (M4 skill) → brief → app builder (Lovable or similar) → live site. Own build, Jul 2026.

🎙 client calls · analyzed into the braintheir words, their pains, their offer📄 one-page brief · pages, copy blocks, style🛠 app builder ships from the brief🌍 live site · client pays for the machine, not hours
GROW · EXAMPLE 2 OF 5

An SEO Content Machine

The brain holds the client’s voice file, offer, and keyword list. A weekly job drafts posts in that voice against those keywords.

You review once, it publishes. Ten drafts, one sitting.

📎 Common agency build: voice + keywords in memory layer, scheduled drafting job, human approval gate before publish.

🗣 voice filehow the client sounds🔑 keyword listwhat people search⏰ weekly job · drafts 10 posts in that voice👤 your review · one sitting · OK / EDIT per post🌍 published · the machine runs weekly
GROW · EXAMPLE 3 OF 5

A Personal Data Dashboard

My training dashboard: watch data lands every night, a job rebuilds one web page, my AI reads the same numbers I see.

Same pattern sells as client work: any data feed → a living page.

📎 Own build: nightly sync from a sports watch API → static page on free hosting → the brain reads it for the weekly review.

⌚ watch → nightly sync · sleep, training, weight📊 my dashboard · rebuilt nightlysleep7h 12mtraining2 × 45mintrend▼ on track🧠 the brain reads it too · Sunday review has real numbers
GROW · EXAMPLE 4 OF 5

This Course Site Itself

Every deck, template and skill file on this site is written, checked and deployed by my second brain, agents in parallel, me approving.

You are looking at the proof right now.

📎 loai.wisdomblend.com: vault → parallel agents (research, build, verify) → git → deploy. Human gate before every publish.

🧠 the vault · course plan, decisions, style rulesresearchagentbuildagentverifyagent👤 my review · the slides you critique, I approve🌍 loai.wisdomblend.com · this page, this slide
GROW · EXAMPLE 5 OF 5

Your Own Product

A game. A micro-SaaS. An MCP server other people install. The brain holds the design doc, user feedback, and roadmap; agents build against it.

The second brain is the co-founder that never forgets.

📎 Pattern across indie builders 2025-26: solo founders shipping games and SaaS with agent teams; the design doc in the brain is the single source of truth.

the braindesign doc · feedback · roadmap🎮 a game📦 micro-SaaS🔌 MCP server💶 recurring €
CLOSING

Next: Module 4

Maintenance: 15 minutes on Sunday, sessions closed clean, two audit prompts a month. Growth: Module 4 starts with the tool list and the first skills you build yourself.

loai.wisdomblend.com · livingoffai.nexus/accountability