Posts

Three Days a Week: What a Rate Limit Actually Costs

Image
In July I wrote that monthly plans beat paying by the token for my development work. I also mentioned getting only three or four working days before hitting the weekly limit. What I understated was how much that limit would start deciding my schedule, and which model reviewed my code. A $20 monthly plan sounds like a month of work. In practice the weekly limit decides how much of that month you actually get, and for me the answer was about three working days. Then I waited. Not for an hour, not until midnight, but for the reset, which is most of a week away when you hit the wall on a Wednesday. Three Days, Then Waiting The plan was Ollama Pro, monthly. Twenty dollars, and on paper that is a bargain next to what the same work costs through an API key. The weekly ceiling is what broke it. I would get through roughly three days of real work and spend the rest of the week unable to do the thing I was paying for. Waiting is worse than it sounds, because the work does not pause polit...
Post a Comment
Read more

Building a Personal Assistant: What I Got Wrong

Image
A couple of weeks ago I wrote that I am the only user of my personal assistant app, so every issue in its tracker is one I filed after watching something fail on my own phone. That sentence is no longer true. My daughter started using it, and she found a limitation I had built in so deeply that I could not see it. Before that, though, some numbers. The app is at version 6.5.0, build 191, with 115 release notes behind it. It is 245 Dart files in lib and 234 test files. The lib/services directory alone holds 124 services. I bring this up because people who ask me about building their own assistant usually think of it as a weekend project - a chat box with an API key behind it. The chat box was the weekend. Everything after that is the reason the thing is actually useful. Three things moved it from a toy to something I open every day: it learns from its own mistakes, it accepts the way people actually talk, and it can look at a photo. Everything else is plumbing, and I will be hone...
Post a Comment
Read more

A Convincing Review of the Wrong Path

Image
This article follows Executor Model, Reviewer Model: Why Cross-Vendor Code Review Actually Works , where I wrote about using a second model to validate review findings. This time, the missing check was not another opinion. It was running the code. I am the only user of my personal assistant app, so every issue in its tracker is one I filed after watching something fail on my phone. One was the fasting timer. I told it in chat that I had started a fast at seven in the evening, and nothing started. I handed that issue and eight others to Claude. It read the source, quoted the right files, and explained why the timer could not fail the way I described. I closed all nine issues that afternoon. A few days later I asked Claude to check again, but this time it had to run the reported inputs. Four verdicts flipped. The second pass took six and a half minutes instead of two and a half. Those extra four minutes recovered four real bugs and found two more I had never noticed to file. T...
Post a Comment
Read more

Executor Model, Reviewer Model: Why Cross-Vendor Code Review Actually Works

Image
I thought one of my projects was in reasonable shape. I had been working in it for months, it was on version 4.24.3, and Claude had been through most of it more than once. Then I pointed Codex GPT-5.6 Sol at it with Ultra High reasoning, asked it to evaluate the codebase and file issues for anything it found, and it came back with 42 new issues in about ten minutes. Some of them were release blocking. I had Claude Opus 4.8 go through all 42 looking for false positives, and it confirmed zero. Every issue was real. That result is what convinced me to stop having one vendor's model check its own family's work. The Setup: One Model Writes, a Different Vendor Reviews The workflow is simple to state. Implement with one model, working from the most critical issues to the least critical and matching model strength to issue criticality: the highest-confidence model takes the most critical work, and the cheaper, faster models take the routine fixes further down the list. Then valid...
Post a Comment
Read more

Claude Keeps Ignoring My AGENTS.md, and I Think I Know Why

Image
In July I asked an agent to push a build of my personal assistant app to my phone. Routine task. I do it most days. It built the release APK, ran flutter install , and my habits, journal, and health logs were gone. Not corrupted, not partially migrated. Gone. flutter install uninstalls the app first when it feels like it, and everything my assistant had learned about me lived in app-private storage with no cloud copy I could count on. The part that still bothers me is not that it happened. It is that the instruction telling it not to do this was already sitting in the repository, in the file the agent claims to read. The Rule Was Already There, in Capital Letters After the first time this happened, I did the obvious thing. I wrote the rule down. The commit is still in my history, and the message reads like an incident report because that is what it was: flutter install uninstalled the app before reinstalling on a real device, wiping local user data (habits, journal, health log...
Post a Comment
Read more

GLM Plans, Fable Validates, and API Keys Are Not for Development

Image
In my last post I compared Codex, Claude, and GLM 5.2 and landed on two $20 monthly plans as the setup that gets the work done. Two weeks in, I changed that setup: Claude moved to the $200 annual plan, which is essentially the $20 monthly plan at a discount, and Ollama moved to the $20 Pro plan. Same shape, better price. This post is a follow-up on two other fronts. First, the workflow itself has firmed up: GLM 5.2 does the assessment and planning on less critical projects, Fable 5 does the assessment on complex or mission-critical projects and validates GLM 5.2's findings, and Sonnet 5 handles general-purpose coding once Fable 5 has set the plan. Second, I ran the experiment I suspected would go badly and paid for AI by the token, once through an OpenAI API key with the GPT-5.6 Terra model and once through Ollama top-up credits. The numbers are worth sharing because the gap is not small. It is the difference between 3-4 days of work and 30 minutes. The Workflow: Who Assesses,...
Post a Comment
Read more

Codex vs Claude vs GLM 5.2: A Practical Comparison

Image
The question I keep getting asked is which AI coding assistant I actually use day to day. The honest answer is that I use more than one, and the reason is that none of them is uniformly the best. Each one has a lane where it clearly wins, and each one has a failure mode that pushes me back to the others. This post is a practical comparison of three tools I have spent real time with: Codex, Claude, and GLM 5.2. I am not interested in benchmark scores or leaderboard positioning. I care about how these tools behave across the full lifecycle of a project: initial scaffolding, iterative refinement, design polish, and long-running reliability. The short version: GLM 5.2 has replaced Codex as my daily driver for implementation work. Codex failed the wife test, which is what pushed me to switch. Claude is strong on design and polish but showed a huge regression after two weeks of use. GLM 5.2 is great at identifying tiny lingering issues and regressions, but the last 24 hours of work showed...
Post a Comment
Read more