The 'Mad Scientist' Prototyping Loop
by Dhanji R. Prasanna • Chief Technology Officer at Block (formerly Square)
CTO overseeing a 3,500+ person engineering organization across Square, Cash App, and Afterpay. A former Google engineer who worked on Google Wave and Google+, he is a key driver behind Block's transformation into an AI-native company through the development of their internal agent, 'Goose'.
🎙️ Episode Context
Dhanji Prasanna details Block's radical transformation from a federated financial services company into a functional, AI-native technology organization. He explains how their internal open-source agent, Goose, saves employees 8-10 hours a week by utilizing the Model Context Protocol (MCP) to give AI 'arms and legs.' The conversation covers the necessity of organizational restructuring to support AI, the fallacy of code quality equating to product success, and the future of autonomous engineering.
Problem It Solves
Prevents 'Boiling the Ocean'—spending massive resources on big ideas that fail (like Google Wave) before validating utility.
Framework Overview
A disciplined approach to innovation that insists on starting with micro-teams solving hyper-specific personal problems before scaling to a platform or product.
🧠 Framework Structure
The 'Cup of Tea' Rule: Don't boil the...
The Personal Itch Test: Founders/buil...
Question Base Assumptions: Before bui...
Controlled Chaos: Allow engineers fre...
When to Use
During Hack Weeks or early-stage product definition to avoid over-engineering solutions for problems that aren't yet understood.
Common Mistakes
Scaling the team size (70-80 engineers) before the product has been validated by real user utility.
Real World Example
Block's Bitcoin integration started as a 3-person Hack Week project (Dhanji, Jack Dorsey, and one engineer) before becoming a major business line.
If you want to make an apple pie from scratch, you have to first invent the universe... narrow your scope to the thing that's in front of you.
— Dhanji R. Prasanna