Dispatches
Careers

Build the artificial employee.


The problems are real, the customers are serious, and the roster below is not a metaphor — it is how we already run.

In person New York City Five days a week Non-negotiable — stated proudly
Engineers at a whiteboard in a New York loft at dusk
01 — Hard problems

The work is genuinely hard. That’s the point.

A collaborative spreadsheet engine in Rust→WASM. 100k-row uploads shipped; million-row scale in progress. Formula evaluation off the UI thread.
Per-cell attribution and undo/redo on CRDT sync. Real-time collaboration built with the creators of the Yjs ecosystem — every agent edit attributable and reversible.
One control plane, every regime. Helm-orchestrated, outbound-only agents delivering the same platform to managed cloud, enterprise VPCs, on-prem, and air-gapped environments.
Autonomy a top-tier regulated firm will sign off on. Reversible actions, perfect attribution, access governance — governance as an engineering problem, not a policy document.
An agent with an inbox, a phone number, and a seat in the meeting. One runtime routing email, Teams, Slack, SMS, voice, and live meetings.
Swap frontier models under live enterprise workloads. Model-agnostic core; customers get new models the day they exist, with zero migrations.
02 — The deal
An elite, small team. No headcount theater.
Ownership is structural — FDEs own more here than at Palantir, with no deployment-strategist buffer between you and the customer.
Ship to the most demanding regulated enterprises in the same week.
Automate your own job with Athena — the product improves because we live in it.
Duty roster — one shared artifact 24 hours
Human
workforce
09:00 – 18:00
Agent
workforce
00:00 – 24:00
Work product
delivered
1 2 3 4
1 02:30 — Athena files draft v3 2 07:02 — analysis delivered 3 09:15 — controller reviews, one edit 4 18:14 — next request sent, by email
03 — A letter from Brendon

We are building the first real artificial employee — something that arrives, learns your procedures, does the work, and can be held accountable for it. That is a systems problem, a product problem, and a trust problem all at once, and we are not going to solve it over Slack.

So we are in person, in New York City, five days a week. Not because we distrust remote work, but because the problems we take on are too entangled to hand off asynchronously — the spreadsheet engineer needs the FDE who needs the person who wrote the governance model, all in the room.

The people who thrive here are unreasonably curious, allergic to hand-waving, and happier shipping to a hard customer than debating in the abstract. If you want to see the frontier of how companies will actually run, this is the place to build it.

— Brendon