Daring to build extraordinary things.
I'm Sam Silva, a software engineer and founder with 17 years of building products, platforms, and AI systems. Now I build systems that check whether AI did what it claims, and I share what I learn along the way.
Now
Systems that check AI.
One question runs through my work: how do you know an AI system did what it says?
Assay
Verification for training AI agents. The grader lives in a separate cloud account the agent can never reach, and every grade is signed.
Evolua
AI code review for security, quality and performance, built on Claude. More than 1,000 pull requests reviewed; bug discovery up 50%.
What I learn
I share what I learn along the way: essays on AI engineering, agents, and making AI systems you can trust.
Recent writing
All posts →Claude's Founder's Playbook, and the agent skill I built from it
What stuck with me from Anthropic's 36-page Founder's Playbook, and how I turned it into an agent skill my agents can load mid-decision — plus 12 more, installable in one command.
- Sep 4, 2026Answer Engine Optimization: how growth changes when AI answers firstWhat AEO is, why AI answers are replacing search clicks, and what I actually did about it: measurement tooling, a demand-driven content pipeline, and community.
- Aug 26, 2026I Fired My AI Company After Seven DaysNine agents, 265 tasks, one deliverable. The post-mortem of a one-person AI company that buried itself in coordination in a week, and the rules the rebuild runs under.
- May 8, 2025Ambient AgentsDiscover how LangChain’s ambient agents and LangGraph move beyond chatbots, automating tasks from email to Slack while keeping humans in control.