
What we've built with eve so far
What we've shipped on Vercel's eve so far: our content-ops agent, how each part of it works, and the things we learnt as a team putting an agent framework into real use.

Thoughts on content systems, frontend architecture, and the tools we use daily. Occasionally opinionated.

AEO, GEO, or just SEO: the playbook is the same. Metadata, JSON-LD, sitemaps, and content negotiation, implemented on a production Next.js site.





Content automation earns its bad name when a model makes decisions that should be rules. Here's where we draw the line, built on the Vercel AI SDK.


Five of our internal skills are now open SKILL.md files, discoverable at /.well-known/agent-skills/index.json. What they do and why we gave them away.





The CMS migration checklist and pre-launch plan we run on every replatform: redirects, sitemaps, JSON-LD parity, OG images, baselines and 30-day post-launch monitoring.


How we structure AEO and SEO on every Sanity build. Fallbacks, JSON-LD, sitemaps, content negotiation, llms.txt, accessibility, and the feedback loop that catches drift.





We pointed Vercel's DeepSec at our own site. Real timings, real token burn, real findings, and why agentic infosec is about to feel normal.




Generate Sanity insertMenu thumbnails automatically with a Claude Code skill. Five to seven minutes per project, no Figma, no manual screenshotting.



Every blog post on this site has a Victorian cat hero image generated by AI. Here's every single one, in full colour, from the earliest originals to today.





Headless Shopify sounds great until you see the bill. We break down when it makes sense, when it doesn't, and what it actually costs across vanilla Shopify, custom themes, Hydrogen, and Next.js builds.

