The accidental AI revolution
summary
An internal developer tool becomes a company-wide agent: Goose autonomous AI agent boosts engineers by 20% more efficient and lets 75% save 8–10 hours/week while engineers build and bundle 100+ MCP servers (Git, Snowflake, Jira, Google Workspace, PagerDuty). Goose is LLM-agnostic, distributed via desktop/CLI and Databricks model hosting, uses short-lived OAuth credentials tied to identity, offers recipe sharing and one-click deploys, and drives adoption through Slack channels, workshops and embedded support.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:14AI-first mindset at Block
- 2:39Developing Goose
- 4:22Engineers saving hours
- 5:48Expanding Goose beyond engineering
- 7:55Scaling Goose to 12k employees
- 10:06Installation and usability challenges
- 13:14Securing MCP servers
- 15:44Authentication and user friction
- 17:31Culture and adoption via Slack
- 18:56Educating employees on AI
- 20:00Goose impact across teams
- 21:27Enhancing Goose features
- 23:35Lessons from scaling Goose
- 26:48Q&A
related talks
12mCodename Goose: An open source local AI agent built on MCP
24mGoose: an open source AI agent
24mBuilding an AI Native Engineering Org
18mBuilding an Autonomous Engineering Org
27mSoftware engineering with GenAI
57mOne Team’s Journey to Using AI Without the Hype
42m10x Engineering: How GenAI Tools Transformed My Team's Productivity
45mPowering AI Agents with Temporal and Kgoose