A few things worth thinking about.
No perfect CV, no buzzword bingo. Just reflect on a couple of questions — and here's what we build, so you know what you'd be part of.
Before you write
A few questions worth sitting with.
No perfect CV needed. What helps us most is understanding how you think.
What do you want from an employer?
Tell us what a great workplace actually looks like for you.
What conditions matter to you?
Hours, location, growth, autonomy — what do you need to do your best work?
What are you genuinely great at?
Show us your strengths. Here, skills and what you've built beat titles and degrees.
What do you actually enjoy?
Passion is what we hire for — what kind of problems make you lose track of time?
How we build
What you'd actually work with.
You don't need to know all of it — you need to want to build with it. We keep the exact stack to ourselves, but here's how we think about the work.
Context engineering, not hype
Most AI failures aren't reasoning failures — they're missing-context failures. We feed our systems exactly the right knowledge at the right moment: one Brain engine carrying thousands of scoped brains (per company, project and entity) with decisions, playbooks, code and history. Getting context right is the craft here.
Human experts + virtual employees
Real specialists paired with AI teammates that read from the Brain before they act and write back what they learn — so the next task starts smarter than the last. We build and run whole virtual-managed companies this way.
Frontier and self-hosted models, used deliberately
We run a mix of frontier and self-hosted models and are opinionated about which job goes where — cost, latency, privacy and quality all decide. Knowing the difference, and caring about it, is exactly the kind of thinking we look for.
Retrieval, memory and knowledge graphs
Everything our systems do is grounded — semantic retrieval, long-term memory, knowledge graphs and event streams — so the right facts surface automatically instead of being prompted in by hand. Grounding beats guessing.
Agentic, event-driven automation
Work moves through playbooks and event-driven flows: something happens, the right agents pick it up, act, and hand off. We design these systems to be observable and safe, not magic black boxes.
Whole companies, designed around human + AI
We don't bolt a chatbot onto old software. We design products, operations, content and support around human + AI collaboration from the ground up — and we measure, evaluate and improve them like real products. If that excites you, you'll fit right in.
Curious? That's the main requirement.
Send a short intro, your CV, maybe a video — and tell us what you'd love to build.