Product
AIP. – Activate Intelligence Platform, the company memory
Activate Intelligence Platform
(around the office, just “ape”)
The context that makes Claude work in your organisation. AIP connects meetings, documents, and decisions into a company memory that your team and its AI can both use.
Why it exists
Most of what an organisation knows evaporates.
Decisions live in someone’s head, context dies in email threads, and every new piece of work starts with the archaeology of finding out what was already known. AIP exists so that stops happening: a company memory that captures what your organisation knows and keeps it working.
For Claude to help with your work it needs context: who the clients are, what was decided last month, and why. AIP brings that knowledge together. The context that helps the AI is the context that helps the people: preparing for a client call, understanding a decision, or taking over a colleague’s project.
We did not build it as a product. We built it because, as a small team working across a dozen clients, we could not afford to forget anything. It has since become the centre of how we operate, and the pattern we deploy for the organisations we work with.
The architecture
Serious software, built in layers.
The substrate
A knowledge graph at the core: people, organisations, projects, meetings, documents, tasks, daily work records, and the relationships between them. Extendable by design – each deployment can add the entity types its work needs, on the same foundation.
The access layer
The graph is exposed through an MCP server, the open standard that lets AI models read and write real systems mid-conversation. Connected AI assistants consult the graph as they work and record what they did. Access is role-based down to the API, private content is envelope-encrypted at rest, and data residency follows the client’s rules – including deployments that stay entirely inside their infrastructure.
The native apps
aip.app runs on macOS and iOS and turns meetings into structured knowledge: recording, diarised transcription, live insight capture, tasks extracted and filed seconds after the end of the meeting. Before a call, it works the other way round – calendar, client graph, recent work, and open tasks assembled into a briefing before the first word is spoken.
The scheduled capture
Memory that depends on people remembering to file things starves. Scheduled services sweep the places where work already happens – GitHub commits, Slack conversations, documents – into the graph on their own, and chase the gaps where a day’s work left no trace. A full-screen dashboard makes the whole thing ambient in the office.
The insights engine
The layer that captures thinking rather than activity: atomic claims extracted from conversations and documents, each with provenance – who said it, where, when. Claims cluster around subjects, disagreements stay alive rather than being flattened, and consensus has to be earned through independent corroboration, not repetition.
Five layers on paper, one memory in practice. The fastest way to judge it is to watch it answer questions about a business it knows.
Who uses it
Same memory, different hats.
We are a small team, so nobody here does just one job. The same person writes code in the morning and sits with a client in the afternoon. The graph does not care which hat you are wearing – it answers the question in front of you.
Writing code
A session starts with what changed last week, which decisions constrain the design, and who is working nearby. The commits flow back into the graph on their own.
Designing
The visual identity, past iterations, and the reasoning behind every rejected direction live in the graph – so new work starts from what was decided, not from a blank canvas.
Facing a client
The briefing assembles itself before the call – calendar, client graph, recent work, open tasks – and the meeting files itself afterwards.
Brainstorming
The insights on a subject become raw material: claims and disagreements pulled up, pushed against, and recombined into ideas nobody had yet. The best thinking starts from what the company already knows, then argues with it.
Writing
Drafting against the graph rather than memory is why our case studies cite what we did, not what we remember doing.
Arriving
Our newest team member spent her first week asking the graph instead of interrupting colleagues.
Running on it
What a year on AIP looks like.
Decisions hold: a choice made in a Monday meeting reaches every AI-assisted piece of work the same afternoon, because everything flows through models connected to the same company memory.
The longer it runs, the more it knows.
One caveat: the automatic layers really are automatic, but recording why a decision went the way it did still takes a sentence of discipline in the flow of work. We think that trade is the whole point.
An asset you tend, not a system you install.
For your organisation
Start with one team.
We start with a six-week pilot on your own AIP instance. Together we choose a piece of work where shared context would help, and agree how to judge the results before we begin.
We set up the instance with your clients, projects, and the initial material we have agreed, write the skill that describes how your firm works, and help the team get started with the app. The pilot includes an onboarding session and a named person on our side for support. The six weeks start when access and the initial material are ready.
As the team works, it adds meetings and documents and asks questions of what they contain. We review useful answers and actual use together at two weeks and at the end, against the questions agreed at the start. You then continue on a licence, or close the pilot with the export and deletion arrangements we agreed.
The pilot has a fixed fee, quoted once we have scoped the team, and we give you an indicative monthly price for continuing before you sign.
We run our company on AIP. We can show you how we use it to answer questions about our own work.