Blog

·11 min read

What Happens to Your AI App at 9am on Monday?

I made 20% of my LLM gateway's calls fail on purpose and it dropped none of 200. Then I simulated 50 clients at 9am on Monday and 20 of them got an error.

Read post

·13 min read

Is Your LLM Ticket Router Good Enough to Ship?

I built a support ticket router with Claude and tested it four ways. It scored about 92% on my 40 hand-picked tickets and about 60% on messy ones, so one accuracy number couldn't tell me whether it was ready to turn on.

Read post

·6 min read

Jev Is Fast Enough to Check Every Tool Call Your Agents Make

TypeSafe shipped Jev in September 2026. It cannot write text. What it does instead is answer a list of questions about a piece of data, with probabilities attached, in about a…

Read post

·3 min read

Finding My Parents in a 24-Year-Old Scanned Voter Roll

It still baffles me how many people use AI today only as a fancy autocomplete, when the thing in front of them can do so much more.

Read post

Serverless vs. Dedicated vs. Self-Hosted LLM Inference: When Self-Hosting Actually Gets Cheaper

I ran the same model three deployment ways on DigitalOcean and measured the real costs, latency, and gotchas to find exactly when self-hosting stops being cheaper than serverless.

Read on DigitalOcean

·6 min read

When Your LLM Gets Retired: What You Actually Need to Do

A while back I got an email from OpenAI. A model I had been running in production, one I had carefully prompted, tested, and tuned, was being deprecated, with 60 days' notice. I…

Read post

·5 min read

I Benchmarked 4 AI Memory Backends So You Don't Have To

Most AI chatbots can't remember what you said five minutes ago. You tell it your dog's name. You share your flight date. You mention a project deadline. Twenty turns later? All…

Read post

·11 min read

Nix: Always On, Always Aware: How I Built My Personal AI

The next wave of AI isn't a better chatbot. It's agents you talk to. Always on, always aware of your world, able to actually do things.

Read post

·9 min read

Let the Model Decide: From Fixed Pipelines to Dynamic Agent Orchestration

A few months ago, I built a travel planner the obvious way. One agent for flights. One for hotels. One for restaurants. One for attractions. Four agents, hardcoded, running on…

Read post

·4 min read

I Got Locked Out of My Own GitHub. An AI Agent Did It.

I was using Claude Code to clean up my GitHub repositories. Updating descriptions for repos that were missing them, batch-organizing things I had been putting off for months. The…

Read post

·4 min read

Agent Loops: The Reason AI Gets Things Done

I asked Claude Code to refactor a module that had grown too large. It read every file in the directory first. Then it planned a sequence of extractions, identified which functions…

Read post

·4 min read

Securing AI Agent Skills Against Prompt Injection

AI agents load skills to gain new capabilities. But every skill is also a set of instructions the model will follow.

Read post

·4 min read

Code Your Videos: Using Remotion and Claude Code Together

I wanted to create a video. It had animated text reveals, code snippets, feature cards sliding in, and transitions timed to a voiceover.

Read post

·7 min read

Carrying Context Across Claude Code Sessions: From Simple to Automated

One thing that becomes obvious once you start using Claude Code seriously on real codebases is this: sessions end, but work does not.

Read post

·2 min read

Running ClawdBot on a DigitalOcean Droplet

ClawdBot has been the talk of the town lately. Like most people, I initially ran it locally on my Mac. It worked well, but over time I started feeling the friction.

Read post

·4 min read

The Rise of the Solo Product Engineer

Coming from a strong agile software development background, I have been thinking a lot about how software development might look in the near future. Not in theory. Not in…

Read post