hev mind

Talk abstract.

First entry in a series on the search tech I actually reach for.

I haven't reviewed a line of code in over a decade — not because I can't read it, but because I learned in 2008 that trust comes from outcomes, not from meddling in output.

Racing is the ultimate expression of engineering, and the privateers — people building their own agent harnesses and competing at the frontier — are where the real innovation in AI is happening in 2026.

Doug Turnbull's Cheat at Search course inspired me to build hev ask — agentic Q&A search for Astro docs sites.

A ⌘K search overlay that's keyword-instant while you type and turns into a bounded agent when you press Enter.

Twenty years of web presence, from a hand-organized folder of files to WordPress to no home at all — and back to a folder of files.

turbopuffer wins the price-performance math but left me asking "now what?"

Search is the fuel for agent context, so solving search for your agent is often the difference between succeeding and failing at a task.

I shut the vibecheck developer preview down over the break — I ran out of GCP credits for the database instance.

Evals were everywhere on the re:Invent expo floor, but the focus felt off — they are becoming observability add-ons rather than a way to reach production.

Episode 27 of Open Source Ready with Brian Douglas and John McBride: what an eval actually is, running structured experiments across model variations, and why cost-efficient design matters more than ever — plus MCP adoption, Claude Skills, and the economi…

A 20-question eval suite on a client MCP integration took nearly eight minutes to run, so every system prompt tweak meant waiting or testing a subset and hoping.

Vectors are transforming the dynamic AI/ML landscape, ushering in a wave of new use cases while shattering long-standing barriers.

Ever wonder what kind of carbon foot print your cloud computing software is producing?

This month has marked the 20 year birthday of the search giant Google, and the 10 year birthday of it’s most successful offspring — the mobile operating system Android.

Having worked on Platform as a Service product for the last few years I have come to realize that Product Management and design for cloud products have a few unique aspects to them that make them different than other products.

I am working with a new team and Program this month and these thoughts have emerged as a theme among discussions with my Product Management peers.

For many operators logging represents an immutable record of their application behavior and a necessary ledger for auditing, security, and…

The observability problem space is big enough to deserve a clearer taxonomy and an honest separation of concerns.

Products expected to just work are hard to research the usual way; nobody offers you feedback on GPS or wifi.

Last year when the Google SRE Team performed their application readiness review of the Cloud Foundry platform they identified two Frequent and Damaging issues related to the log transport system Loggregator.

Heuristics are rules of thumb, and here’s why they are so effective at managing bias — and some practical examples for Product Managers lifted from the world of alpinism.

Product managers should be as comfortable at the command line as they are with sticky notes and 2x2s.

Software has to be easy to operate, scale, and maintain, not just solve the user's problem.

Most PMs learn to read patterns out of user needs, then stall when those patterns meet architecture decisions.

Should designs be attached?

With the publication of Google’s Site Reliability Engineering book many software teams are working to improve their ops practice and start…

Whatever the methodology, the ability to scale small agile teams within a larger organization is crucial to succeeding in competitive…