Own the engine: the live agent loop, the data pipelines that feed it, and the measurement that keeps it honest in production. What you’ll do The live agent loop: streaming conversations, tool use, and recommendation logic serving real shoppers on real stores. Catalog intelligence: ingestion and enrichment that turn a brand's products, policies, and reviews into something an agent can sell from. The agent-ready data layer that ChatGPT, Perplexity, and shopping agents read, reason over, and recommend. Honest measurement: A/B infrastructure, attribution, and the eval harnesses that catch a production LLM the moment it drifts.
Startups
Bron 2026-10-08