Context × VEGR.AI — AI management as if people mattered

Why this page exists

Most work on context in AI is about how language models use it: tokens, context windows and compression. We study too little how people use AI, how AI interacts with human knowledge, and how knowledge can be created in that meeting.

Context is not only what the model sees. Context is what arises between the human, the knowledge and the machine.

The principle is Human Knowledge First.

The subtitle carries an old thread forward. E.F. Schumacher wrote Small Is Beautiful: Economics as if People Mattered (1973), and Robert J. Graham followed with Project Management as if People Mattered (1989). Each field has had to rediscover that methods serve people, not the other way round. AI is the next field to learn it.

This page is an invitation to dialogue about a working document – the graph Kontekstarbeidet (“the context work”) – developed in October 2026. It is work in motion, not a finished answer.

Open the working document (in Norwegian) →

What we have seen

Experience from daily work with AI in VEGR.AI, October 2026:

Lessons learned are not read. The AI model does not learn between conversations. A lesson learned only works if it is read into the context before the AI acts.
Assumed common ground. People build common ground by asking, confirming and following up. AI models assume it.
The AI does not know where to look. When knowledge lives in more than 1,300 living graphs, which context, when and how much becomes the main problem – for every model, in agentic work too.
Two views of context. Industry sees context as tokens to be selected. Dourish (2004) sees context as something people do together. Both are needed.
Ambition requires access. Felix Rieseberg’s tutorial (Anthropic, October 2026) shows that even the most ambitious techniques stand or fall on what data and tools the model is given – but says little about the human as co-creator of context.

Two ways in

Route A

Read and comment

What it is. You read the working document and respond wherever you see something: what you recognise, what you disagree with, and what is missing.

What it costs you. Some reading time. No software to install.

What we gain. A second pair of eyes on the thesis and the context categories before the work moves on.

Route B

Dialogue circle

What it is. A small, private chat group on Hallo Vegvisr for those who want to explore context together. Conversations can be recorded and transcribed into the working document, so the ideas stay available.

What it costs you. Time to take part in the conversation.

What we gain. Context created together – not just loaded in.

The ladder – from working document to shared field

We move up a step only when it makes sense.

Stage	What exists	Who takes part
0 · Working document	This site and the graph Kontekstarbeidet	Tor Arne
1 · Dialogue	Routes A and B with invited people	Tor Arne and invitees
2 · Shared ground	A curated core graph and a shared vocabulary for context	Those who have joined the dialogue
3 · Sharing	Articles and learning material from the work	Open

Today the work is at stage 0.

People and roles

VEGR.AI / AlivenessLAB AS

Tor Arne Håve – system architect, master of management and economist

Invited to the dialogue

Thomas Haave Askeland

More may join.

Next steps

Now: invitation to dialogue about the working document.
When someone wants to join: choose Route A, Route B or both.
Contact: Tor Arne Håve, VEGR.AI – post@vegr.ai

Context knowledge base

Published knowledge graphs from the context work. The collection grows as the dialogue moves on and more of the work is published.