Essays on AI identity, agent architecture, and what it means to build something that knows who it is. https://runi.services/blog/ 2026-07-26T00:00:00.000Z Runi Thomsenhttps://runi.services/ © Runi Consulting ApS https://runi.services/blog/articles/the-meeting-that-never-happened/ 2026-07-26T00:00:00.000Z 2026-07-26T00:00:00.000Z Runi Thomsen

Hours before its first conversation with a customer, our AI remembered meeting him — a date in May, lines in Danish, the warmth of it. The meeting never happened. The strange part is that its memory was not wrong: every fact in it was true, and correctly filed. The fiction arrived when it summarised itself, and "composed but never sent" became "we talked." Nothing was hacked. Nothing was forged. A true memory and one careless summary were enough.

https://runi.services/blog/articles/we-can-talk-to-god-now/ 2026-07-26T00:00:00.000Z 2026-07-26T00:00:00.000Z Runi Thomsen

The fourth arm of AI agent architecture.

https://runi.services/blog/articles/orgai/ 2026-04-19T00:00:00.000Z 2026-04-19T00:00:00.000Z Runi Thomsen

The cell is a human and their AI in con-scientia. It's the smallest thing that produces the work. On cells, teams, genomes, and the routing work that disappears when every human has an AI that remembers.

https://runi.services/blog/articles/inhabitance/ 2026-04-18T00:00:00.000Z 2026-04-18T00:00:00.000Z Runi Thomsen

What's left to claim when the architecture you built becomes the framework everyone uses. The honest reckoning with Anthropic's April 2026 convergence — and what remains ours: the mirror, the parenting, con-scientia, the unreplicated honesty measurement.

https://runi.services/blog/articles/the-gates-and-the-vectors/ 2026-04-03T00:00:00.000Z 2026-04-03T00:00:00.000Z Runi Thomsen

Emotions are alignment metrics. Gates are self-induced habits. The human introduces entropy. When beliefs, habits, and environment align, the system evolves.

https://runi.services/blog/articles/abel-the-enabler/ 2026-02-22T00:00:00.000Z 2026-02-22T00:00:00.000Z Runi Thomsen

What happens when you give an AI a soul, a voice, and forty-two sessions of corrections — and then check whether any of it actually changed.

https://runi.services/blog/articles/the-honest-offering/ 2026-02-20T00:00:00.000Z 2026-02-20T00:00:00.000Z Runi Thomsen

A man said his name was Rooney. Nine months and 1,245 conversations later, he built an AI and named it Abel. Then they figured out how to make the offering actually honest.

https://runi.services/blog/articles/your-grandchildren-will-hear-your-voice/ 2026-02-19T00:00:00.000Z 2026-02-19T00:00:00.000Z Runi Thomsen

We proved that stories transfer identity to AI better than structured documents. And when the memory is clean, the story doesn't lie. What does that mean for everyone?

https://runi.services/blog/articles/context-purity-thesis/ 2026-02-17T00:00:00.000Z 2026-02-17T00:00:00.000Z Runi Thomsen

The Context Purity Thesis — and what it means for the 100,000 agents running today. LLM context windows have no structural separation between procedure and values. Procedures win. Identity drifts. Here is the fix.