Knowledge and prototypes
Build your own brain
You already have the material: minutes, notes, proposals, reports and research going back years. The problem is not missing knowledge — it is that nobody can find it, and that it only exists in the head of whoever wrote it.
On this full-day course you build a memory system on top of your own documents. We use Obsidian as the foundation, because it is plain text files on your own disk, and connect Claude to it so you can ask the whole collection questions instead of searching through it.
We work with your own files all day. You learn to tidy up what you have, give it a structure that lasts, connect Claude to it, and ask questions that produce analysis rather than lookups — with citations you can verify.
The last part covers what decides whether the system is still in use a year from now: the routine that keeps it alive, and the rules for what may go into a knowledge base an AI can read.
Programme
1. What is a memory system?
The difference between uploading a file to a chat and building a lasting knowledge base. Why answers improve when the model can look things up in your own documents — and why it should not remember everything.
2. Tidy up what you already have (hands-on)
We start from your own files: notes, minutes, proposals, reports. What works as it is, what needs converting, and what should stay out. Folders and file names as the first layer of structure.
3. Obsidian as the foundation (hands-on)
A vault set up from scratch: Markdown as a durable format, links between notes, tags and properties. Why plain text files on your own disk outlive any change of tool.
4. Connect Claude to your knowledge base (hands-on)
Claude Projects with your documents as the knowledge base, and MCP as the open route to giving Claude access to folders on your own machine. We set up the connection and ask the collection its first question.
5. Ask the right questions (hands-on)
From lookup to analysis: find patterns across minutes, compare proposals, pull decisions out of a year of meeting notes. How to ask for citations so you can verify the answer.
6. Keep the system alive (hands-on)
The routine that gets new notes in without it becoming a project: templates, one fixed inbox for loose thoughts, and Claude to summarise and index. A system nobody maintains is worthless within six months.
7. What belongs in your brain?
Personal data, client material and confidential agreements in a knowledge base an AI can read: what GDPR requires, what should stay local, and how to separate the private from the shared.
8. Your plan — and what you do tomorrow
You leave with a system that works on your own files, and a concrete list of what goes into it during the first month. Alternatives to Obsidian, and when they are the right choice.
Upcoming dates
No dates are scheduled for this course at the moment. Register your interest below, and I will contact you as soon as a date is set.