Build the product surface for AI agents that do real insurance work - the tools people use to create, run, and supervise agents, and the experiences they trust to act on the results.

Ship across the whole surface: agent authoring, a chat assistant that reads documents and takes actions, human-in-the-loop supervision, and multimodal experiences across documents, voice, and the web.

Own features end to end, idea to shipped, across the stack (React/TypeScript front end, Python back end).

Set the quality and taste bar, and mentor.

Representative Projects

The Universal Assistant - one assistant on top of everything a company runs on. It reads their documents, pulls from the systems they already use, reasons across all of it, and takes the action - so work that used to bounce between five tools and three people just gets done. You'd build the interface and the intelligence that make that feel effortless, and trustworthy enough to rely on.

Agents that run for hours, not seconds - long-running, human-in-the-loop agents that wait on an email, make a phone call, and know exactly when to hand control back to a person. You'd design what it even means to supervise an agent mid-flight.

Multimodal by default - agents that work across documents, voice, and the web, in one product. Building product experiences that stay coherent whether the input is a PDF, a phone call, or a live website.

Staff Product Engineer

Responsibilities

Qualification

Required

Benefits

Company

Further AI

More roles at Further AI