🤖 Code junkie and many other things. ✨ Tech & AI | 💪 Fitness & Nutrition | 📚 Politics & History | 🦄 Gay Stuff | 🌴 Miami, FL | 🇲🇰🇺🇸
Miami, FL
Joined December 2015
- Tweets3.8K
- Following1.4K
- Followers1.1K
- Likes2.4K
If you're not tired of Opus 5.5 motion videos yet, here's one more. I had it make a promo film for my new app, and I honestly think it came out great.
If you use both ChatGPT and Claude, you've probably copied answers between them to see what the other thinks. Errol does that for you. It carries each reply between the two Mac apps until they agree they're done. Free and open source: errol.chat
Now that Fable 5 is out, what's your "habit tracker" project? I don't know who needs this, by here's mine: longhand, neural handwriting synthesis for web and mobile. A real model writing real ink, not a font. Type a line, watch it written, export stills or animations. trylonghand.com
Not a font. Same words, new seed, new handwriting every time. There's an SDK too: TypeScript for the web, Swift and Kotlin for mobile. github.com/tmarkovski/longha…
Announcing a small project I've working on: Argus, an open-source workflow viewer for DBOS Transact apps.
I tried DBOS Transact and was sold instantly. Three client projects later, I'm still hooked: durable workflows in a Postgres you already have. No dedicated clusters, no cloud-vendor state machines, no k8s sidecars.
(@DBOS_Inc is co-founded by Postgres creator Mike Stonebraker, whose grander ambition is to one day replace Linux with a database-as-kernel OS. Another story.)
For production-grade workflow management (recovery, alerting, retention, team controls), DBOS has Conductor. It wires into your app config and talks to a managed service. I wanted a dev one-liner:
uvx dbos-argus --db-url "postgresql://..."
MIT, read-only, self-hosted. localhost:8090. If you're playing with DBOS, I'd love to hear what's broken. 🫶
github.com/tmarkovski/dbos-a…
I think much more critically now than I did before AI because the information is much more available to me.
I'm going to call this right now. We are going to have a large population with absolutely no critical thinking skills if they blindly trust AI for everything.
We have all already seen it.
They don't validate outputs. They don't really understand anything. They just ask questions, it looks good, and they go with it.
There are going to be huge issues in every company as this continues over the years. The amount of technical debt and knowledge gaps are going to be insane.
So much opportunity if you actually know what you're doing.
Tomislav Markovski retweeted
IDEMIA 🤝 Trinsic
Excited to announce this partnership to accelerate mobile driver’s license (mDL) adoption 🚀
Now @trinsic_id is the go-to acceptance layer for mDLs powered by IDEMIA, giving people more places to securely use their mDL online.
Read more in replies 👇
I don’t know who needs to hear this, but if you’re using ChatGPTs Voice Mode to have a conversation with your faithful assistant, know it’s using GPT-4o, a very old model. It’s not updated to the latest.
OpenClaw is just the 2026 version of recompiling your Red Hat kernel with custom flags you didn’t understand, except now you’re burning through $200 in tokens to build a glorified cron job that sends you a Slack reminder to drink water.
I think the real issue isn't compiled languages vs interpreted languages - it's whether AI tools can see your code. They don't struggle with compilation, they struggle with ecosystems that ship opaque binaries without source or symbols, and what David's pointing out in .NET applies just as much to any stack that treats its artifacts like black boxes. Did you hear that C/C++?
.NET is at a slight disadvantage with coding agents since the distirbution mechanism is binaries (dlls). I've been experimenting with a decompile skill using the ilspy command line tool.
You can install it here:
github.com/davidfowl/dotnet-…
#dotnet #agents
I just converted a PNG image to SVG for free with Adobe online converter and it did it without asking me to pay, sign-in, throttle, watermark, or anything else promotional. I can't shake this - I still feel there's a catch.
Tomislav Markovski retweeted
Claude is great at coding. Like really, really good, compared to using a regular LLM like Gemini.
There's only one issue with it. Because it never sees the full code of the app but uses grep search for relevant code snippets, it's myopic.
If grep returns a fragment of code similar to the bug description, it often doesn't look further and fixes an irrelevant part of the app or answers a question based on these fragments found by grep.
So, as the codebase grows, it becomes important for the user to know the codebase. Otherwise, Claude will reinvent the bicycle over and over again, creating duplicate implementations for the same functionalities in different places in the app.
This issue is probably fixable with additional finetuning, but right now this is how it works.
How many bottom bars does it take to present information on iOS? I'm shocked and disappointed at the direction (or lack thereof) the Apple team has taken with this.
Florida is facing one of the largest tourism collapses in modern US history. It was not caused by a hurricane or a pandemic. It came from a political rupture with Canada. In under two years, more than 280,000 tourism jobs vanished and over $52B was erased. Canadian visitors once brought $38–42B a year into Florida. That flow collapsed by more than half in a single winter, gutting snowbird dependent counties. 🧵👇
The most unsettling part is permanence. Planning models now assume only ~30% of former Canadian winter spending returns over the next decade, leaving a $35–40B annual hole. Florida is projected to lose 160,000 seasonal residents per year long term. This was not retaliation. It was recalculation. When trust disappears, markets withdraw.
Follow me for more random thoughts.