About
AI proposes.
Humans decide.
We build the layer that makes that work.
What we believe
Three things we hold to.
Constans suggests. Humans decide.
Every answer the system gives is a proposal. Every decision is confirmed by a human, who owns the call. AI doesn't get authority. Humans get help.
Knowledge compounds when it's governed.
Information without provenance is overhead. The same answer reinvented in four corners of the org is waste. We build the layer that makes knowledge traceable, versioned, and reusable - so what's resolved once stays resolved.
Mechanism, not magic.
Constans does not "get smarter" by hand-waving. It captures missing knowledge, validates it with humans, and makes it permanent through a defined process. Every claim cites its source. Nothing important happens silently.
Why this exists
It started by paying attention to what AI couldn't do.
Constans began as an exploration into where AI tools fall short - what ChatGPT and Claude can't do, no matter how large the model gets. As the work moved toward consultancy use cases, a pattern kept surfacing: knowledge doesn't go missing because the answers don't exist. It goes missing because the answers aren't traceable, versioned, or owned.
Constans is what that realisation turned into.
Who builds it
A small team led by Colin Montgomery.
Nine years at McKinsey & Company delivering digital products to large enterprises. Building software since the late 1980s.
What was always missing was a system that captured the decisions teams actually made, governed them, and made them reliably retrievable. Constans is that system.
"Senior leaders write astutely about why governance keeps failing them. Almost none are building toward fixing it. Constans is what fixing it looks like."
Introduce Constans into your organisation.
Reduce repeated work. Improve consistency. Ensure decisions hold over time.