Personal AI OS

Your second brain, productised — with the AI tools that read it.

For founders, owners and senior leaders whose ChatGPT and Claude conversations all start with the same five minutes of context-setting. We build you a personal operating system — a custom CLAUDE.md, a memory layer that captures decisions and preferences, a daily-briefing workflow, and the AI tools wired into it — so the next conversation picks up where the last one ended.

Book a fit call See the two builds
Starter £1,995 fixed
Full £4,995 fixed
Delivery in 2–4 weeks
UK-hosted UK GDPR-aligned
Who it’s for

Built for people whose time costs more than the build.

This is not a tool. It’s a personalised operating system. We don’t sell it widely — the build only makes sense if your time is worth more than the engagement fee inside the first 90 days. Three buyer profiles consistently see that payback.

Profile 01

Founder / MD

You run the business. You start every AI conversation with the same context dump. You make 5–10 strategic calls a day and you wish you had a sharper sparring partner with full context.

  • Decision log that survives sessions
  • Daily briefing across all live concerns
  • AI that knows your business, your team, your customers
Profile 02

Operations director

You hold the operational picture for a 20–200 person business. The cost is in re-explaining current state to every new conversation. A working second brain ends that.

  • Always-current operational state
  • Per-project context dossiers
  • Inbox triage tuned to your priorities
Profile 03

Senior leader in a regulated org

You work in NHS, central government, defence-adjacent or finance. You need an AI workflow that respects classification boundaries and leaves an audit trail. The standard tools don’t.

  • Classification-aware memory layer
  • UK-hosted only, per-engagement DPA
  • SC-cleared lead on regulated work
What’s inside

Six layers, in plain files you own.

Personal AI OS is deliberately built from plain Markdown files, not a proprietary app. Everything we build is portable, readable, and yours — if you ever want to leave, you take the files and the AI tools they configure still work.

CLAUDE.md

Your top-level instruction set. Defaults, behaviour, locked decisions, communication style. Read at the start of every session.

MEMORY.md

Decision log + preference register. Corrections, things you wish you hadn’t done, the “always do X / never do Y” rules.

People & projects graph

Who’s who, what’s active, what’s on hold. Backlinks. Updated weekly via the daily-briefing workflow.

Daily briefing

A 5-minute morning summary across your meetings, inbox, project status changes, and items that need your decision. Yours alone — no shared inbox.

Inbox triage

Full-build only. Auto-classifies inbound into respond-now / respond-later / archive / delegate, with draft replies for the first two buckets.

Document drafting

Full-build only. Trained on your past correspondence + decisions. Drafts board updates, investor letters, memos, follow-up notes in your voice.

Pricing

Two builds. Both fixed-price.

We don’t do a monthly retainer for Personal AI OS — it’s a one-off build with a 90-day check-in. After that, you own it. If you want ongoing tuning we’ll quote it separately, but most clients run with the build for 6–12 months before needing anything.

Starter

The foundation layer.

£1,995fixed · 2 weeks

The core four layers: CLAUDE.md, MEMORY.md, people & projects graph, daily briefing. The right starting point for most leaders — pays back inside 90 days at typical professional billing rates.

  • Two weeks, mostly remote, 3–4 short interviews
  • Custom CLAUDE.md tuned to your voice + preferences
  • MEMORY.md decision log seeded with your current context
  • People & projects graph (50–200 entities typical)
  • Daily-briefing workflow live by week two
  • 90-day check-in to tune the layer
Start with Starter

Both tiers exclude VAT. Optional KnowledgeForge retainer at £2,500/month if you want us to keep the context layer current as your business changes — not bundled, deliberately.

How it works

From fit call to working OS — five steps.

Fit call

30-minute call. We’ll be honest if this isn’t the right product for you yet. Most people benefit; not everyone needs the full build.

Discovery

Two short interviews. We map your decisions, your repeating patterns, your terminology, the people and projects in your orbit.

Build

We assemble the six layers (or four for Starter). You sign off on the CLAUDE.md before anything goes live.

Train

One session to walk you through how to use it — the “correct twice = update memory” pattern, the daily review ritual, the boundary on what to delegate.

90-day check-in

We come back at day 90 to tune what’s drifted and decide whether the optional retainer makes sense. Most clients don’t need it.

Frequently asked

The questions we get on the first call.

Why would I pay for this when I can build a CLAUDE.md myself?

You can — and if you have the time and patience to iterate it over several months until it’s actually load-bearing, please do. Most leaders we work with try this for two or three weekends, end up with a thin file that doesn’t survive contact with their actual work, and stop. The build pays for itself if you would otherwise have spent four-plus weekends on it — which, at most senior pay rates, you would.

Will the OS work with Claude only? What about ChatGPT and Gemini?

The CLAUDE.md is named for Claude because it’s the model we recommend as primary — the long context window and the system-prompt behaviour are well-suited to a personal OS pattern. But the files are plain Markdown; we ship parallel adapter notes for ChatGPT (system prompt + custom instructions) and Gemini (Gem configuration) so you can run the same OS across multiple tools. Most clients end up using two or three models in different contexts.

How private is this? My CLAUDE.md will contain sensitive context.

The files live wherever you choose — locally on your machine, in your own iCloud or Google Drive, or in a vault on your own server. Launchpad retains no copy after the engagement closes. The build is covered by a per-engagement DPA and, for regulated-sector clients (NHS, MOD-adjacent, financial services, central government), the engagement runs under appropriate clearance and a data-classification map.

What we’ll explicitly never do: store your CLAUDE.md on a Launchpad-controlled server post-engagement, share it with any third party, or use it for training. The IP is yours from day one.

How is this different from KnowledgeForge?

KnowledgeForge is for businesses — a context layer your AI agents and team read. Personal AI OS is for individuals — a context layer that one named person reads from and writes to. The deliverable on the page is similar (Markdown files, memory layer, governance), but the audience and shape are different. Many full-build Personal AI OS clients also commission a separate KnowledgeForge for the company they run; they’re complementary, not overlapping.

What if I don’t already use Obsidian or Notion?

Fine. Most of our Personal AI OS clients don’t. We’ll usually recommend Obsidian (free, local-first, plain Markdown) as the home for the files, set it up for you as part of the build, and walk you through the daily review pattern. If you’re already on Notion, we’ll work in Notion. If you’d rather keep the files in plain folders on your laptop with no app at all, that works too.

Will I become dependent on this? What happens if I leave Launchpad?

Everything is in plain Markdown files you own. If you parted ways with us tomorrow, the OS would keep running — Claude, ChatGPT and Gemini would all keep reading the same files. The engagement is deliberately one-off precisely so the dependency you end up with is on the files, not on us. We’d much rather you didn’t need to call us again.

Stop re-explaining yourself to your AI.

Book a 30-minute fit call. We’ll talk through your week and tell you honestly whether the Starter, the Full, or none of the above is the right answer for you.

Book a fit call