AI Engineer – Autonomous Agents

HelloCounsel builds AI agents that run pre-litigation for US personal injury firms: intake, case opening, medical records, insurance claims, client communication. They work across phone, email, fax and the firm's case management system. Live at real firms, $3.2M seed. Next: run 80% of a pre-lit case with no human in the loop. THE ROLE You build the agents that do the work after the first call: open the case, request medical records, chase insurance claim numbers, keep the client updated. They run for days or weeks on each case, across email, fax, portals and the firm's case management system. They have to be right, because a missed deadline or a wrong record can cost someone their case. You own them end to end. Nobody hands you a spec. WHAT YOU'LL DO Build long-running agents: planning, tool use, state and memory across days, retries, and a clean hand-off to a human when stuck Plug into case management systems, email, fax and portals: through APIs where they exist, and around them where they don't Build the evals, test sets and monitoring that prove an agent is right, and catch it fast when it isn't Read documents at scale: medical records, bills, police reports, insurer letters Ship every week. Track task success per case. Kill what doesn't move it. YOU'RE A FIT IF You've shipped LLM agents that take real actions in production, not just answer questions You've built evals and can show how you knew your system worked You're a strong engineer: Python or TypeScript, APIs, queues, databases, and debugging messy distributed systems You go deep on a new domain fast. No legal background needed. YOU'RE NOT A FIT IF Your agent work is notebooks and demos You need a spec before you start HOW WE HIRE A short written problem, a call with the founder, then a small build: make a working AI agent. We judge how you get past obstacles, not polish. In office, Bengaluru.