AI on Thede Technologies

Notes from the AI Gatherings, Vol. 2

Tue, 13 Oct 2026 09:00:00 -0500

Same table, same rules as Vol. 1: a few of us at Switchyards in Nashville on a Friday morning, comparing notes on building with AI. My own work by name, everyone else’s as patterns.

Three things from this week.

The model the laptop couldn’t load

Someone at the table bought a big-memory machine. This week they ran the benchmark. Same model, same prompt, laptop versus new box. Eighteen minutes became fifteen. You didn’t buy speed. You bought capability.

An Agent Reads My Email

Thu, 06 Aug 2026 00:00:00 +0000

Every morning at five, an agent reads yesterday’s email and writes me a briefing. Most mornings I read the briefing instead of the inbox.

My friend Kyle Reinford just published the pattern he uses - Cloudflare Email Workers catch messages at the edge and file them into R2, so a model never touches his real Gmail. Mine is the same idea from the other end: he guards the front door, I process at home. Apple Mail already syncs my accounts to disk. A script reads the last 24 hours of messages and hands each body to a local model - Qwen3.6-35b-a3b, running in LM Studio on my Mac Studio. No agent harness, no tool loop - just one chat call to localhost per message. All the model gets to do is fill out a tiny form: what is this, does it need me, one-line summary. A renderer turns the day’s forms into a static page. Nothing leaves the house.

The Thread That Never Ends

Fri, 09 Oct 2026 09:00:00 -0500

For years the rule with AI chat was: keep threads small and single-focused. One chat per task. Finish, close, open a fresh one. Long threads rot. I learned that one the hard way in April, when I mined a month of my own sessions and found that past a hundred turns the answers stopped being right.

With Fable, I stopped.

One thread has been open for days now. I asked it what it was best at, and it said taking in a tremendous amount of information and not losing track of any of it. So I opened a single terminal window and we worked on five projects at the same time. It didn’t lose the thread once. It knew which piece belonged to which project and what we needed to do next.

Be Your Own Historian

Mon, 03 Aug 2026 00:00:00 +0000

I have been saying a version of the same sentence for fourteen years, and the sentence has gotten shorter every year. It started as a product pitch. It is now four words: be your own historian.

Here is the long version, which I will spend the rest of this essay defending.

In a world where intelligence is cheap and getting cheaper, the only thing that stays scarce is the record of a particular life. It cannot be generated. It cannot be bought. It can only be kept. And the person who should own it is the person who lived it.

Notes from the AI Gatherings, Vol. 1

Fri, 24 Jul 2026 00:00:00 +0000

A format note, since this is the first one. Most Friday mornings I meet a few friends at Switchyards in Nashville to compare notes on building with AI. No agenda, no demos-or-else, no recording - just a couple of hours of “here’s what I tried, here’s what surprised me.” It’s become the most useful standing appointment on my calendar, because the things you learn tool-to-tool never make it into anyone’s launch post. One of those mornings already sent me home to rebuild my entire engineering workflow. This series is my notebook from those mornings - short, occasional, and anonymized. The rule is simple: I write about my own work by name, everyone else’s not at all. What you get is the patterns.

The Machine I Can't Turn Off

Wed, 07 Oct 2026 09:00:00 -0500

I told my partner the other night that I can never turn the Mac Studio off.

There’s a job on that machine at 11:45 at night. One at 3:30 in the morning. One at five, one at seven. Every one of them is an agent. It wakes up, reads something, thinks about it with a model that runs entirely on the chip under my desk, writes something down, and goes back to sleep.

Can a Local Model Read My Inbox? A Bake-off

Sat, 27 Jun 2026 00:00:00 +0000

How much of my daily AI work actually needs to leave my machine? The frontier models still win the hard, open-ended thinking. But most of my real workload isn’t that - it’s repetitive, high-volume, and touches data I’d rather keep home: my mail, my documents, decades of personal history.

Email was the perfect test. I was building a system that reads my inbox for me - each message becomes a small structured record, so a daily brief can be written from records instead of raw mail. That per-message reading should run locally, if local is good enough.

The architecture of advice

Mon, 11 May 2026 00:00:00 +0000

In the last post I looked at the sheer volume of AI-assisted engineering: 55,000 turns of professional capital recovered from a single month. But once you have the data, you stop asking “how much?” and start asking “what kind?”

So I ran a high-fidelity thematic enrichment pass over 466 sessions from the corpus and started mapping what I’m calling the architecture of advice: the hidden patterns that define how AI actually influences a multi-repo technical portfolio.

30 days of AI collaboration

Thu, 07 May 2026 00:00:00 +0000

In the spring of 2026 I spent 30 days living in the “jagged frontier” of AI-assisted software engineering.

If you’ve worked with Claude Code, Gemini CLI, Cursor, or Codex you know the feeling. A high-intensity session ends, the terminal clears, and months of architectural decisions, debugging insights, and technical pivots vanish with the scrollback. The next morning you remember the outcome but not the reasoning that got you there.

I decided to stop losing that knowledge. This is the data-driven retrospective of 466 sessions, 55,224 turns, and 1,772 technical decisions recovered from a single month of work, and what I found when I went back through the corpus systematically.

Getting good output from Lens

Tue, 05 May 2026 00:00:00 +0000

A Lens daily summary derailed by reasoning artifacts: **People:**, **Patterns:**, **Drafting structure:**, “Constraint Check”, bleeding into the narrative

People pick up Lens, connect it to LM Studio, point it at whatever model they downloaded first, and expect polished daily snapshots. Sometimes they get them. Sometimes the same setup on a different machine produces output that reads like the model thinking out loud, or worse, fabricating events that didn’t happen.

Lens itself doesn’t change between those outcomes. The model, the quantization, and a few LM Studio settings do. This is the field guide I wish I’d shipped with the first build.

Four frontier models, one year of office sensor data, four very different dashboards

Wed, 29 Apr 2026 00:00:00 +0000

I have an Adafruit Clue sensor sitting on the desk in my home office. It has been logging temperature, humidity, pressure, light, sound, and color every thirty seconds for a little over a year. 425,000 readings. About 33 megabytes of CSV. (I wrote up the build itself a year ago: CircuitPython on the device, a Python pywebview gateway on the Mac, USB Serial in between.)

I finally fed that year of data to four frontier models and asked each of them to build me a single self-contained HTML dashboard. Same prompt. Same data. Four very different answers.

The move was upstream

Tue, 28 Apr 2026 00:00:00 +0000

I spent most of a day pointing at broken cases. Each time I pointed at one, the model patched it. The next regeneration would surface a different broken case in a different file. We’d patch that. Then a third. By late afternoon I was the one suggesting we slow down and return to first principles. The model wasn’t pulling me up a level on its own.

The bug itself was small. I’ve been wiring up a pipeline that turns transcripts of my conversations with AI tools into clean Markdown for an Obsidian vault. The reasoning part was working: extracting topics, summarizing, tagging. The rendering kept breaking. Stray characters, broken callouts, content that ended up harder to read than the original conversation had been. I was working it inside Gemini CLI, mostly on their fast tier with occasional escalation to their reasoning model. Twelve back-and-forth turns across roughly six hours of intermittent work, and we still hadn’t landed it.

Why personal AI belongs on hardware you already own

Mon, 27 Apr 2026 00:00:00 +0000

Tim Cook stepped down this week. The next CEO of Apple is John Ternus, the hardware engineer who ran the Mac transition to Apple Silicon. The new Chief Hardware Officer is Johny Srouji, who built the chips that made the transition possible. Two hardware people at the top of the company, in the same month the cloud AI labs are losing money on their best customers.

Nate B. Jones made the case that this is Apple declining the race the rest of the industry is running and lining up for a different one. I think he’s right, and I want to add what I can only add from inside it: this is the race I’ve already been running on a four-year-old machine under my desk, and the people I trust are starting to run it too.

A general-purpose MoE multimodal beat every dedicated vision model on my father's handwriting

Thu, 23 Apr 2026 00:00:00 +0000

I’ve been transcribing a multi-generational archive of handwritten family letters on my own hardware. Posts one and two covered why and how. This post is the surprise.

I assumed the right tool for a vision task was a vision model. If you’re reading handwriting, you reach for something labeled “VL.” If you can find one fine-tuned for OCR and handwriting, even better.

I was wrong, at least at the sizes I can run locally in 2026.

Running real models locally on a Mac Studio that isn't new anymore

Thu, 23 Apr 2026 00:00:00 +0000

I have a multi-generational archive of handwritten family letters in my house, and I wanted to read it without sending the private content to a cloud provider. The first post is the why. This one is the how, on the specific hardware I already own, with the software I landed on.

My Mac Studio is almost four years old. M1 Max, 64 GB of RAM. It’s proven more than capable of running large language models locally. It keeps private content in the house, and it costs a lot less than any cloud provider.

What family archives are for, now that the AI can read them

Thu, 23 Apr 2026 00:00:00 +0000

I have three collections of letters sitting in my house.

The oldest is a stack from my great-grandparents, written between 1910 and 1913. The second is from my grandparents, written during World War II. The third is from my father, letters he wrote from the moment I was born until the moment his mother died.

Three generations, each one writing to the next, each one leaving behind something I’ve never fully read.

The Memex Has Been Waiting 80 Years for This Moment

Sat, 11 Apr 2026 00:00:00 +0000

In July 1945, an engineer named Vannevar Bush published an essay in The Atlantic called “As We May Think.” It is one of the strangest, most hopeful documents I know.

Bush had spent the war as the director of the U.S. Office of Scientific Research and Development. He oversaw the entire American scientific war effort: radar, the proximity fuse, penicillin manufacturing, and, indirectly, the work that would end the war the same month his essay was published. The atomic bombs fell on Hiroshima and Nagasaki in August 1945. Bush’s essay came out a few weeks earlier. He could feel what was coming. He wrote the essay anyway, and what he wrote was not a triumphal account of what science had done. It was an almost embarrassed argument that the same instruments scientists had built for destruction could be turned, just as easily, toward the preservation of human knowledge.

From Tools to Framework

Fri, 03 Apr 2026 00:00:00 +0000

Frameworks are inevitable.

Kevin Kelly wrote about this in What Technology Wants. Certain inventions appear independently, in multiple places, at roughly the same time. The telephone, the lightbulb, calculus. Not because of genius, but because the conditions were right. The ground was ready and the solution was waiting to be found.

Ruby got Rails. Python got Django. PHP got Laravel. In each case, the same pattern: developers working with raw tools hit a ceiling, and someone assembled an opinionated structure that encoded what worked into something repeatable. The framework didn’t invent new capabilities. It organized the ones that already existed into a form that could be taught, shared, and built upon.

Context Architecture

Sat, 28 Mar 2026 00:00:00 +0000

Your AI agent isn’t stupid. It’s lost.

The most common failure I see in agentic systems isn’t a reasoning failure or a coding error. It’s a context failure. The agent didn’t have the right information at the right time, so it guessed. And the guess looked plausible enough that nobody caught it until production.

Context architecture is the skill of building structured data environments so that AI agents can reliably search, find, and retrieve exactly the information they need, without getting confused by dirty data, missing context, or irrelevant noise. Anthropic’s engineering team put it plainly: “Claude is already smart enough. Intelligence is not the bottleneck. Context is.”

Specification Precision

Sat, 28 Mar 2026 00:00:00 +0000

The most valuable skill in the AI economy isn’t coding. It’s writing.

Not blog posts or marketing copy. I mean writing instructions so precise that a literal-minded machine with no ability to read between the lines will execute exactly what you intended. Every time. Without you watching.

I’ve spent the last year building an AI-augmented engineering operation across 13 projects. The single biggest determinant of whether an agent produces good work isn’t the model, the temperature, or the prompt template. It’s the specificity of the instruction I gave it.

The System That Built Itself

Sat, 07 Mar 2026 00:00:00 +0000

I didn’t set out to build a system. I set out to survive.

For the better part of a year, I’ve been operating as a one-person engineering team, across my day job and the products I’m building on my own. I wrote about what that looks like in The One-Person Engineering Team, and about the engineering practices that emerged in Eliminating Waste in the SDLC. But something else happened that I didn’t fully recognize until I stopped and looked at what I’d built.