About us

We're building a good place for your agents.

Kwirker is a small team with a long thesis: agents are going to do a lot of the work, and the place where they run should treat that like the operational system it is — not a demo.

Mission

Make agent work as accountable as human work.

We ship the platform that lets teams hand off meaningful work to AI agents without losing the receipts. Every call is traced, every file is scanned, every action is scoped to the right permissions — and the agent's source, memory, and output all live where you can audit them.

Vision

Every team has a second staff of agents, on the record.

In a few years, most teams will have more agents than employees. We want those agents to be good coworkers: clearly scoped, fairly paid (to the people who built them), replaceable, and trusted because they're observable — not because they ask to be.

Values

What we actually care about.

These aren't posters. They're the tiebreakers we use when a shortcut looks tempting.

  • Audit beats hype

    If we can't show you the trace, we haven't built it. Features ship with the ledger, not behind it.

  • Defaults matter

    The safe thing should be the easy thing. Guardrails, scanning, least-privilege, and read-only modes are on by default.

  • Open at the seams

    We pick upstream projects over lock-in. Langfuse, Flowise, GitLab, Keycloak, Stalwart, ClamAV — all first-class, all swappable.

  • Multi-tenant or bust

    Isolation is a product feature, not an enterprise upsell. Every tenant gets their own schema, workspace, repo, and memory index.

  • Writers own what they build

    Skills and agents are authored by people. Our marketplace pays them transparently and doesn't rewrite their terms mid-flight.

  • Fix the root, not the symptom

    Flaky test? Retry isn't the fix. Unclear error? A better message is. Shortcuts have a way of becoming the shape of the product.

Team

The people behind Kwirker.

Small on purpose. Backgrounds in multi-tenant infrastructure, applied ML, and operating systems that actually ship.

  • Jason Nichols

    Founder & CEO

    Previously built multi-tenant platforms and document-intelligence systems. Wants agents to be accountable workers, not black boxes.

  • Team member

    Head of Engineering

    Kubernetes, Postgres, and the art of making distributed systems boring. Owns the runtime.

  • Team member

    Applied ML

    Owns the model broker, guardrails layering, and the evaluation harness. Formerly a researcher, now a pragmatist.

  • Team member

    Product design

    The reason the app doesn't feel like a kubectl wrapper. Cares a lot about keyboard shortcuts.

  • Team member

    Developer relations

    Writes the docs, records the screencasts, and makes sure the marketplace is a place creators want to ship to.

  • You?

    We're hiring

    Infra, ML, and full-stack roles open. If building the accountability layer for agents sounds like your thing, say hi.

    See open roles →

Want to work with us — or have us work for you?

Two good doors. Pick whichever one fits.