Talk to Will about a team setupAsk, get a cited answer.
Bring a question. Get the answer with the source note and the date it was written, or a straight "I don't know". It never fills the gap with a guess.
Team Memory MCP Server
Give your whole team one brain that answers with receipts.
Your best context lives in one person's head, a Slack thread, a doc no one opens. This turns the notes your team already keeps into one shared brain they can ask. Every answer comes back cited to the note and the date, and each person only sees what you let them. Connect once in Claude or ChatGPT. No code.
Cited answers or an honest "I don't know". Per-person access, default-deny. Pairs with the Intelligent Growth methods and the Marketing Brain.
Every answer is citedSource note and date, or it tells you it does not know.
Each person sees only their partDefault is deny. You grant access one folder at a time.
Works where the team worksConnect Claude, Claude Code, or an eligible ChatGPT workspace.
The problem underneath
A shared drive is not a shared brain.
The notes pile up, but nobody can find the decision from three months ago. So people re-ask, re-explain, or just guess. Every new starter takes weeks to load.
The context already exists. It is just not answerable, and it is not safe to open the whole thing to everyone.
What it does
Four things your team gets on day one.
No new app to live in. It runs inside the tools your team already opens.
Start any task already briefed.
Name an account, a project or a topic. Get a cited summary of everything the team knows, newest first, before you write the email or join the call.
The team feeds it in one line.
Say "record this" and it saves into that person's zone with a date. The brain gets sharper every day and nobody has to maintain it.
Each person sees only their part.
Give everyone a scope. Private, financial and sensitive notes stay invisible to anyone without access, not just locked behind a warning.
What changes
Same question. A different kind of answer.
The point is not a longer reply. It is an answer you can trust in front of a customer, tied to the note it came from, scoped to the person asking.
A blank chat or a shared drive
- A confident guess
- No source you can check
- Often out of date
- Everyone sees everything, or nobody does
Your team brain
- A cited answer
- The note and the date behind it
- An honest "I don't know" when unsure
- Scoped to who is asking
How it works
Three steps. Then the team just asks.
Point it at your brain
Connect it to the notes your team already keeps: your Marketing Brain folders, an Obsidian vault, a git repo of markdown. No migration, no new place to write.
Give each person a scope
One line per teammate for what they can see. Default is deny, so a new person sees nothing until you grant them a folder.
Everyone connects once
They add the secure connection in Claude or ChatGPT and start asking. Answers come back cited and scoped to them.
What you get
The memory, and the guardrails to share it.
A shared brain is only useful if you can open it to the team without opening everything.
Cited answers
Team check-ins
One-line capture
Weekly roll-ups
Per-person access
Invisible private zones
Audit log of every read
Your data on your network
Why it is safe to share
Default is deny. Out of scope means invisible, not just locked.
Access is enforced in the server, not by asking the model to behave. If you have not granted someone a folder, those notes never show up in their search, their lists or their roll-ups. They cannot see that the file exists.
The rest holds the line too: it stays inside the vault, skips dotfiles and secrets, redacts anything key-shaped, and writes every read to an audit log tagged with who asked. It is never on the public internet, only reachable over your private connection.
Why I built it
Teams keep losing the one thing that made them good: what they know.
I have spent years watching hard-won context walk out the door with whoever wrote it down, or sit in a folder nobody can answer from.
A model is easy to swap. Your team's knowledge is not. So I built the layer that makes that knowledge answerable and safe to share, without handing the whole vault to everyone.
It runs on your own notes, on your own machine. You keep the brain. The team just gets to ask it.
Where it fits
The brain, the methods, and the layer that shares them.
Three parts that work on their own and better together. You can start with any one.
Marketing Brain
The context your team owns: company, customer and systems knowledge, as plain files you keep. Build it in the cohort or start free.
See the Marketing BrainIntelligent Growth methods
The subscribed library of product marketing methods that runs inside Claude and ChatGPT and applies to the context you provide.
See the methodsTeam Memory MCP
This. It turns that brain into a shared memory the whole team can ask, cited and access-scoped. One person's brain becomes everyone's.
Get early accessGet started
Start with the free brain. Add the team layer when you are ready.
Brain Starter
A$0
The free folder to stand up your own brain, plus the guide to connect it. Build it solo, then bring the team.
- The starter brain, files you own
- Setup guide for Claude and ChatGPT
- Cited recall over your own notes
- Upgrade to the team layer any time
One email with the starter and the setup guide. No spam, unsubscribe any time.
Team Brain
Early access
Your brain, served to the whole team with per-person scopes, cited answers and an audit trail. Set up with you, priced per team.
- One shared memory across the team
- Per-person default-deny access
- Check-ins, capture and weekly roll-ups
- Set up in about fifteen minutes with you
Founding teams hear first, with setup help included.
Choose your start
Stand it up yourself, or have me set it up with your team.
Self-serve
Grab the free starter, point it at your notes and run your first cited recall. When you want the team on it, add scopes and hand each person a connection.
Get the free starterSet up with you
If you would rather not touch the plumbing, message me. I will point it at your vault, write the scopes with you, and run the access test against your sensitive folders so you can watch it get denied.
Message Will on WhatsAppThe practical bits
Questions worth asking.
How is this different from the Intelligent Growth methods?
The methods are the how-to for marketing work: positioning, messaging, launches. This is your team's own context, made answerable and access-scoped. Use both together. The methods do the thinking, the team brain remembers what your company knows.
Is this the Marketing Brain?
It is the Marketing Brain made shareable. The cohort teaches you to build the brain as files you own. The Team Memory MCP serves those files to your whole team, with cited answers and per-person access, so one person's brain becomes everyone's.
Where does my data live?
On your own machine or network, reached over your private connection. It never goes to the public internet, and nothing trains on it.
Can one person see another person's notes?
Only if you grant it. Default is deny, and out-of-scope notes never appear in that person's search, lists or roll-ups. They cannot even see that the file exists.
Does it ever make things up?
No. It answers with the note and the date, or it says it does not know. That is the whole reason you can trust it in front of a customer.
What do I connect it to?
Any folder of markdown notes. The Marketing Brain starter, an Obsidian vault, or a git repo all work. If your notes are somewhere else, message me and we will sort it.
Which clients does it work in?
Claude, Claude Code, and eligible ChatGPT Business, Enterprise and Edu workspaces with custom MCP enabled. Each person connects once.
How do I get it for my team?
Get early access above or message me. Setup is about fifteen minutes: point it at your vault, write one line per person for what they can see, and hand out a connection each.
See it in action
Watch it refuse to leak, live.
That refusal is the whole product. Leave your email and I will send the starter and set up a look.