How Block is becoming the most AI-native enterprise in the world | Dhanji R. Prasanna

Dhanji Prasanna presents Block’s AI strategy through Goose, open protocols, functional organization, and small experiments. He argues that AI expands productivity but does not replace human judgment about which problems matter, and that leaders should use the tools themselves while preserving openness and meaningful work.

AI-native teams are changing how software is built: some use background agents and continuous processes, while others gain less from tools because of legacy systems. The discussion treats reported productivity gains as an evolving baseline rather than a fixed promise.

Block’s most ambitious AI experiments move beyond responding to prompts. One engineer lets Goose observe his screen, conversations, and email so it can infer useful next steps, build a first version, and open a pull request without being given a detailed prompt. The example is deliberately experimental and intrusive, but it shows the intended direction: an agent that anticipates work rather than waiting for a human to describe every action.