How Block is becoming the most AI-native enterprise in the world | Dhanji R. Prasanna
Block CTO Dhanji Prasanna explains how Goose, Block’s open-source AI agent, is saving employees 8-10 hours weekly and why non-technical teams are benefiting most.
- Block’s AI-forward engineering teams self-report 8-10 hours saved per week; company-wide estimate is 20-25% of manual hours saved.
- Non-technical teams (risk, legal, support) show the most surprising AI productivity gains, not engineering.
- Prasanna wrote an AI manifesto to Jack Dorsey while working part-time after paternity leave; it led directly to his CTO promotion.
- Block’s single biggest productivity unlock was organizational: switching from GM/silo structure to a unified functional org (all engineers under one leader).
- Goose is a free, open-source, MCP-based AI agent (Mac/Windows/Linux) that can orchestrate Salesforce, Snowflake, Tableau, and write its own MCP extensions.
- Code quality has almost nothing to do with product success — YouTube beat Google Video despite storing videos as blobs in MySQL on a slow Python stack.
- LLMs run idle overnight and on weekends; Prasanna believes they should be building features autonomously in anticipation of what teams want.
2025-10-26 · Watch on YouTube