7 Things Your AI Coworker Needs - Philipp Schmid, Google DeepMind

Philipp Schmid of Google DeepMind shows how to build an autonomous AI coworker for finance with Gemini and the Gemini API's Managed Agents. He breaks a coworker down into seven needs (a brain, a computer, playbooks, tools, credentials, house rules and a schedule) and builds each one: skills and files in the agent's own sandbox, built-in tools and MCP servers, credentials swapped in by a network proxy so a prompt injection can't steal them, hooks for approvals, and cron triggers. He finishes with a live demo of an agent that audits a portfolio and updates a Google Sheet on a schedule. Chapters: 0:00 Intro 0:57 What every coworker needs 1:54 What's new in Gemini 4 4:29 Managed Agents: Gemini with its own computer 5:31 Building Bug, our AI coworker 6:30 Skills, playbooks and files 8:47 Tools and MCP servers 9:57 Credentials an agent can't leak 12:55 Hooks for house rules 13:36 Schedules and triggers 14:27 Demo: an automated portfolio audit 16:38 Recap Philipp Schmid https://deepmind.google Join us in SF for the second annual AIE CODE summit: https://aie.codes our last flagship summit of 2026. Applications live now!