← All Episodes
Based on Lenny's Podcast data
Lenny's Knowledge Sketch

How Block Became the Most
AI-Native Enterprise

Dhanji R. Prasanna
CTO, Block (formerly Square)
OCT 26 2025
The Bet

Block Rewrote Itself
for the AI Era

AI INTEGRATION DEPTH
"We didn't add AI features. We asked: if we were building Block today, how would AI change every system, every workflow, every product decision?"
  • Block's AI thesis: AI-native architecture, not AI-bolted-on
  • Payments, commerce, Cash App, Tidal — all rebuilt with AI layers
  • The developer bet: empower every Block engineer to build with AI
  • Dhanji's rule: if a task takes more than 2 hours and is repetitive, it should be automated
Framework

Block's AI-Native Architecture

AI TOOLSAI WORKFLOWSAI PRODUCTS
80%
of Block engineers use AI daily
40%
reduction in code review time
faster feature deployment
  • Layer 1: AI tools for engineers (Cursor, Copilot, custom Block tools)
  • Layer 2: AI workflows (automated testing, PR review, incident response)
  • Layer 3: AI products (Cash App AI insights, Square AI analytics)
  • The philosophy: AI should eliminate the most frustrating parts of every job
Dhanji's architecture principleDon't build AI that does the job for engineers. Build AI that removes the parts of the job engineers hate.
Enterprise AI Reality

What Block Learned

  • Works: Code generation, test automation, documentation, on-call triage
  • Hard: Multi-system reasoning, financial compliance automation, trust and safety
  • Unexpected: AI adoption is highest among senior engineers, not junior
  • Culture shift: "AI shame" — hiding AI tool usage — took 6 months to reverse
The AI shame problem

Dhanji found engineers hiding their AI usage for fear of being judged. Creating psychological safety around AI tools took deliberate culture work.

The senior engineer paradox

Senior engineers adopted AI fastest because they could evaluate output quality and knew where to apply it.

Playbook

Build an AI-Native Company

  • Mandate AI tool usage before you try to persuade — adoption needs a push, not just a pull
  • Measure AI usage and output quality separately — usage is a leading indicator, not the goal
  • Create internal AI guilds — communities of practice spread best practices 10× faster
  • Give engineers budget and time to experiment — mandate exploration, not just utilization
The Block standardDhanji sets a company standard: any repetitive engineering task that happens more than weekly should be automated within a quarter.
Contrarian

Enterprise AI Myths Block Disproves

AI is for tech companies onlyINSTEAD →Block serves the unbanked, small businesses, and everyday consumers. AI improves their products too.
Senior engineers resist AIINSTEAD →Senior engineers adopt AI fastest. They have the judgment to use it well.
AI slows down compliance-heavy industriesINSTEAD →AI speeds up compliance when designed for it. Block uses AI to make compliance workflows faster, not to bypass them.
You need a dedicated AI teamINSTEAD →AI is everyone's job. Dedicated teams create silos; distributed adoption creates transformation.
𝕏︎ X / Twitterin LinkedIn📸 Instagram🔗 Copy link
0:00