Jonathan Wondrusch (Block): Raising Ambition in the Singularity
Summary
Jonathan, a Staff Plus Engineer at Block, discusses how AI adoption at Block, shares his journey from skepticism to becoming an early adopter of AI agents. He details using AI for large-scale code migrations and API design, emphasizing the need for human judgment and rigorous evaluation. The conversation highlights balancing AI's speed with collaborative processes to avoid low-leverage work and ensure high-quality outcomes.
Topics discussed
Introduction and guest background
The impact of AI on engineering velocity and scale
Personal journey from skepticism to AI adoption
Driving team adoption and the shift to collaboration
Human touchpoints and cross-functional feedback
The final 5%: Refinement and production readiness
Orchestrating evals and human-in-the-loop validation
Adversarial testing and code review bottlenecks
Raising ambition and avoiding low-hanging fruit
System 1 vs System 2 thinking in AI-assisted work
API design, GraphQL, and domain modeling
Using LLMs as judges for design patterns
Design reviews and system invariants
Outsourcing thinking and verifying AI outputs
Documentation, voice, and learning retention
Agent configuration files and scheduled agents
Current coding habits and future of software building
Listen ad-free on Castria