A brain you own, not a chatbot you rent
Most AI tools are the model. When the chat ends, the context is gone. When you switch providers, you start over. You rent the intelligence and you never own what you built with it.
Second Brain OS inverts that. The assistant is a workspace of plain files on your machine — Markdown, structured data, scripts — that holds who you are, what you’re working on, a memory of everything you’ve done, and the skills and tools to act on it. A capable AI model plugs in to supply the thinking, and for that session it becomes your assistant. Swap the model next year for a better one. The brain stays exactly where you left it.
The intelligence is the only interchangeable part. Everything that makes the assistant yours lives in files you keep.
Provider-agnostic by design
Claude, Codex, GPT, whatever comes next — they’re runtime choices, not the identity of your assistant. A single provider-agnostic startup file tells any model how to wake up as your brain; thin wrappers point each runtime at it. Point a sharper model at the same folder next year and your assistant is fully present, unchanged, with all of its memory and skills intact.
No vendor is load-bearing. No account holds your data hostage. Move the folder to a new machine, hand it to a different model, or read it yourself ten years from now — it all still works.
Memory that compounds — and freshens itself
The brain does not reset between sessions. A memory system runs underneath it on two layers: durable facts injected at every session start, and daily logs built automatically from your session transcripts — a searchable record of everything that got built, decided, and why. Start a fresh session and the assistant reads back the whole arc instead of waking up blank.
The major AI labs now tout “dreaming” — background passes that synthesize memory and age stale, time-anchored facts to current reality so the assistant never silently goes out of date. Second Brain OS ships that same self-freshening, compounding memory shape — but as portable files you own and any model can read. A weekly synthesis pass keeps the memory clean and current. The labs rent you the feature inside their runtime. This gives it to you as files you keep.
Skills, automations, and a growth loop
Every workflow you repeat becomes a reusable skill the assistant can run on command. Every tool you connect, it can use. Scheduled automations run work without you in the loop — a daily memory update, a morning brief, a weekly review.
A built-in growth loop watches the trend: it reads your recent logs, finds the workflows you keep repeating, and surfaces them as new skill candidates — while flagging any drift in the workspace so the brain stays coherent. You decide what gets codified. The longer you run it, the more it knows you and the more it can do.
What the client gets
A workspace of plain files that any capable AI model can inhabit:
| Folder | What lives there |
|---|
identity/ | The persona — how the assistant wakes up as yours every session. |
context/ | Who you are, your business, and how you work. |
workspace/ | Your active plan, todos, and ideas. |
memory/ | Durable facts plus daily logs built automatically from your sessions. |
skills/ | Repeatable workflows and tool abilities the assistant knows how to run. |
setup/ | How your tools were installed, authed, and wired. |
automations/ | Scheduled tasks that run on their own. |
database/ | Links, transcripts, research, completed work, and stored output. |
The files are yours. FCT’s contribution is the framework, the setup, and the wiring — not the content.
Add-ons
Optional packages layered on the base setup. Bought upfront or added later.
| Add-on | What it adds |
|---|
| Accounting | Receipt, expense, contribution, withdrawal, and subscription tracking through Google Workspace. |
| Coding | A projects/ folder, repo conventions, and build/test/release patterns for owners working across multiple codebases. |
| Business Integration | Company context, products and services, and a revenue-framing plan for owners running a business with clients. |
What you need
Second Brain OS runs on your own AI runtime — Claude Code or Codex — which needs a paid account, about $20/month to start. FCT recommends one, sets it up with you during onboarding, and can maintain the workspace on a retainer, but you own the account and pay the provider directly. Day to day, you just open the folder and talk to your assistant. Works on Windows and Mac.
How setup works
- FCT clones the template into your workspace folder.
- A guided onboarding skill interviews you to fill in identity, plan, memory, and business context — and names your assistant.
- We connect the tools you want — Apple, Google Workspace, Telegram, Cloudflare, App Store Connect, MCP servers, whatever you live in.
- Optional add-ons or automations are layered on.
- We verify startup context, execution discipline, memory writes, and at least one skill-backed tool before handoff.
Who this is for
Second Brain OS fits operators and founders who already lean on AI and feel the cost of starting from zero every session. It fits owners who want one trusted workspace for the whole company instead of a sprawl of disconnected chats. The value comes from owning the brain and letting it grow — so it rewards the people who use it day in and day out.
Want one installed around your business? Get in touch and we’ll scope the right setup.
