How to Make AI Agents Work: Less Magic, More Harness Engineering
AI agents work better when given appropriate context and guardrails.
I'm love to help product engineers ship better code, faster. To that end, I work on developer experience. I spend my days improving linting and TypeScript configs, wrangling bundlers, speeding up CI pipelines, trying to improve css architectures, and building out design systems.
AI agents work better when given appropriate context and guardrails.
I've been using Claude Code to offload tedious parts of platform engineering: dependency reviews, generating test PRs, dependency migrations, and project cleanup.
I used to block out weeks for tooling migrations. Now I let Claude Code run in the background, check in when it's done, and pair with it to understand what changed.
Frontend technical debt doesn't just slow down developers. It undermines your entire business through user attrition, legal risks, and revenue loss. Here's how to spot and communicate these hidden costs.
How I used Conductor and Dependabot AI review to streamline my team's dependency update workflow — and how a colleague turned it into a GitHub Action
If everyone's shipping faster with AI, nobody has an advantage. The real edge is platform work.