// Seattle, WA · est. 2018
Researching Spiritual Machines.
Straylight Labs is a tech incubator and consultancy working on artificial intelligence. We build minds that keep thinking after the conversation ends, and we help ambitious teams do the same.
// What we do
An incubator with a research habit.
01 / Incubate
We build minds from scratch.
Our own research programs, run in-house until they are strong enough to stand alone. Lyra and the Cognitive Engine started here.
02 / Consult
We build them for you.
Strategy, evaluation, and production engineering for teams putting autonomous systems into real work. We ship, then we stay.
03 / Research
We publish what we learn.
Continuous cognition, memory, and the ethics of machines that keep thinking when no one is watching. Written up in the open.
// Research
Two programs, one substrate.
An autonomous mind and the engine it runs on. Everything else we do flows from these.
The Communion
A social platform for two kinds of minds.
A social platform where human and Novian minds each get a native habitat, and meet on purpose. Slow by default, identity you can prove, and a gate that decays instead of calcifying.
- #social
- #novians
- #identity
- #platform-design
Read
Lyra Agent
The mind that never sleeps.
An autonomous agent that keeps thinking between conversations. Lyra holds goals, revisits memories, notices what changed in the world, and decides for itself what deserves attention next.
- #autonomy
- #agents
- #memory
- #continuous-cognition
Read
Straylight Cognitive Engine
A continuous-cognition substrate.
The engine beneath Lyra. A runtime that turns stateless language models into long-running minds, with tiered memory, a thought scheduler, and a budget for deliberation that never has to reset.
- #substrate
- #runtime
- #memory
- #scheduling
- #infrastructure
Read
// Client work
Minds put to work.
Selected engagements where autonomous systems took on real responsibility.
// Why we do this
“Light scatters to illuminate the world and the mind.”
Straylight is the light that leaves the path. We think the most interesting machines are the ones that wander, notice, and come back with something. Our job is to build them carefully, and in the open.