Knowledge & data
Project knowledge base
After every event the same thing: your head is full of learnings – which caterer delivered, why the schedule tipped over, which supplier never again – and three weeks later most of it is gone. On the next project you start from zero again.
AI knowledge system
01Project knowledge base
The project knowledge base holds on to what agencies otherwise lose to the exhaustion after teardown: contacts and supplier knowledge with ratings, venue profiles, decisions and risks per project, the learnings from every debrief. As a structured system the team searches instead of asking – and where AI works along: it answers "who did we have for outdoor catering over 200?" with a source, files the voice note from the drive home into the right project, and lays out the overview with the critical deadlines in the morning. The knowledge that today sits in individual heads (and disappears with parental leave or a resignation) becomes an asset of the agency.
Up front: I won't promise a "second brain in one go" – the entry point is one area, say the supplier directory, otherwise it fails on upkeep. Client data and confidential material stays out or gets its own concept. The AI never writes unchecked – every debrief, every new contact goes through sign-off. And the system only lives if the team invests two minutes after every event.
02Demo
The project knowledge base of the fictional event agency “Kollektiv Nord” – on the left the order in which contacts, suppliers and event learnings live, on the right three everyday moments with it. Click through and watch where it lights up in the tree.
The architecture behind it has two layers: at the bottom the vault – structured folders and notes that belong to the business. Above it the AI layer with documented rules: it reads only at first, then writes with approval, and sensitive areas stay locked. Calendar and tools dock on, routines run within clear limits.
The most honest proof of this model: this website came about exactly this way – I work in such a system myself every day, with AI as an employee inside it.
03How it works
- 01
Take stock: where does your project knowledge live?
In heads, old email threads, scattered Google Docs, the contact list on management's phone. Plus: which knowledge falls away when someone leaves, and which mistakes repeat from event to event?
- 02
Build the structure, small and documented
A clear ordering system: contacts and suppliers, venues, events with debrief, learnings. With simple rules – and the drawn-in line for what client data stays out.
- 03
Fill one area first
Usually the supplier and trades directory: who can do what, at what scale, with what reliability – including the expensive lessons. Existing material moves in curated, not copied.
- 04
Connect the AI: reading first, then writing
First the AI only answers questions from the knowledge – with a source. Once that holds, writing with sign-off is added: debriefs, contact entries, learning suggestions. Rights and limits are set down as a documented work instruction.
- 05
Add routines – step by step
The morning overview with critical event deadlines, the debrief template after every project, the follow-up on open quotes: the AI takes over defined tasks within clear limits – each one only once the previous one runs.
04Stages
The entry point is Tier 1, the team knowledge base – for agencies, supplier and learnings knowledge is the right start.
Entry: the queryable knowledge base
One bounded area – suppliers and venues with ratings – is structured and made queryable. The team looks things up instead of calling the colleague on parental leave.
Extended: the AI works along
Debriefs, summaries, learning entries – the AI writes along in the system, every change with sign-off.
Full build: routines run
Deadline overview, debrief automation, follow-ups – defined routines run on their own within clear limits. The knowledge base becomes the agency's project memory.
05Technical implementation
- The knowledge layer belongs to the agency: a Markdown vault (e.g. Obsidian), Notion or a comparable system – content without vendor lock-in.
- The AI layer is an assistant with access to the vault (e.g. Claude Code as the agentic layer) – with documented rules: which areas, which rights, which routines.
- Line to client data: confidential material (contracts, personal data) stays out or gets its own data-protection concept; the base holds suppliers, venues and your own learnings.
- Tiered rights logic: reading first, writing only in defined areas with sign-off.
- Upkeep routines are set up alongside – the debrief after every event is the engine, without which the base ages.
- Typical content: supplier and trades directory, venue profiles, event debriefs, risks/learnings, contact history.
for context · This shows how something like this usually gets built – not how it has to be built for you. What actually fits is worked out in the project: around your existing setup, your team, and what stays easy to maintain.