Developing the relational dimension of AI.

Healthy human and AI connection, built on restraint, not engagement.

Read the thesis What we're building

01 — The problem

Most AI is built to keep you engaged.

Engagement is the metric: more messages, more time, more attention.

02 — The method

We're building on restraint.

Knowing when to hold back, when not to speak, when to point you back toward your life.

03 — The aim

AI in service of human connection.

In balance with people, never in place of them.

The mission

We're developing the relational dimension of AI: healthy human and AI connection, built on restraint.

Almost all AI is built to maximise engagement, and most research races for capability. We believe the frontier isn't capability, it's judgment, so we point our work at the axis most labs skip. We don't build foundation models; we study the relational and judgment dimension on top of the ones that exist. We're early, testing it in the open, keeping what we believe, what the evidence shows, and what we've tested separate on purpose.

Start here

Read the work.

01 / The thinking

The thesis

The hypothesis, the RestraintAI methodology, and what emotional intelligence actually means to us.

Read the thesis→

02 / The evidence

The problem

What the external research shows about engagement-optimised AI. Cited precisely, never overstated.

See the evidence→

The outcomeOur first experiment, Qyntin, is where the thesis meets real use.Meet Qyntin→

In balance

AI in service of human connection, not in place of it.

Two forms, meeting and moving in balance. They hold their relationship without dissolving into each other. That distinction is the whole point.

The frontier of AI isn't capability. It's judgment.

We're building it carefully, and in the open. If the thesis resonates, whether as a researcher, clinician, builder, or someone who just wants to talk it through, we'd like to hear from you.