Alex Smagin

Playing with technology for the better half of my life.

[Alex Smagin @asmagin_]

@tobi Shamelessly stealing "company moves at the speed of its slowest secret."

I constantly catch myself trying to align language, break silos to make projects work.

Companies' T-cells will work extra shifts fighting transparency but approach is worth trying!

[Alex Smagin @asmagin_]

@figma Complex diagrams in mermaid create monsters with auto-layout.
@figma how does it work with FigJam? is LLM responsible for layout composition?

[Alex Smagin @asmagin_]

@PawelHuryn Polishing LLM results and trying to fine-tune behavior or perf, doing it with natural language instead of code, is still coding - the difference is translation to machine code is abstracted further away.

[Alex Smagin @asmagin_]

@dessaigne @RaphaelDabadie @Foaster_ai McK or not - organizations need experience that they do not have internally (and scapegoat if things are not working well). Setting "north star", backcasting, gap analysis are not new.

Leveraging agents as a consulting automation subscription is interesting packaging though.

[Alex Smagin @asmagin_]

@sytaylor Is it model or actual data for training?

[Alex Smagin @asmagin_]

@ecommerceshares So basically, they started from CLI and moving towards IDE designed around agentic workflows... In parallel with claims that IDE would be dead by EOY.

[hattie @oh_that_hat]

There's a fruit fly walking around right now that was never born.

@eonsys just released a video where they took a real fly's connectome — the wiring diagram of its brain — and simulated it. Dropped it into a virtual body. It started walking. Grooming. Feeding. Doing what flies do.

Nobody taught it to walk. No training data, no gradient descent toward fly-like behavior. This is the opposite of how AI works. They rebuilt the mind from the inside, neuron by neuron, and behavior just... emerged. It's the first time a biological organism has been recreated not by modeling what it does, but by modeling what it is.

A human brain is 6 OOM more neurons. That's a scaling problem, something we've gotten very good at solving. So what happens when we have a working copy of the human mind?

[Alex Smagin @asmagin_]

"program.md" sounds very similar to "program.cs" from my dotnet background.

The more we trying to get better results from AI the closer we get to development best practices and project management approaches.

We just writing now at the higher level of abstraction.

[Garry Tan @garrytan]

Karpathy just open-sourced autoresearch.

One GPU. 100 ML experiments. Overnight. You never touch the code — just write a Markdown file.

The bottleneck isn't compute. It's your program.md.

https://t.co/zcR77tlOgO

[chiefofautism @chiefofautism]

someone connected LIVING BRAIN CELLS to an LLM

Cortical Labs grew 200,000 human neurons in a lab and kept them alive on a silicon chip, they taught the neurons to play Pong, then DOOM

now someone wired them into a LLM... real brain cells firing electrical impulses to choose every token the AI generates

you can see which channels were stimulated, the feedback from the neurons in choosing that letter or word

[Alex Smagin @asmagin_]

@JorgeCastilloPr Can you configure that in @kilocode ?

[Alex Smagin @asmagin_]

@rohanpaul_ai It is strange to say that "SaaS is Dead" but "Our SaaS is different".

- Software is not only coding.
- AI cost is not zero, even if you have less people.

AI makes building easier and cheaper, but doesn't magically replace ERP or CRM.

[Wildminder @wildmindai]

17,000 tokens per second!! Read that again!
LLM is hard-wired directly into silicon. no HBM, no liquid cooling, just raw specialized hardware. 10x faster and 20x cheaper than a B200.
the "waiting for the LLM to think" era is dead. Code generates at the speed of human thought.
Transition from brute-force GPU clusters to actual AI appliances.
https://t.co/Bf6DH7Q6Uf

[Stitch by Google @stitchbygoogle]

🚨 Our #1 most requested feature is here:

You can now export designs from any Stitch agent directly to Figma as editable layers.

Vibe Design is perfect for exploring many ideas in minutes. But sometimes, you need that final layer of polish.

Now you can move seamlessly between vibe design and polish.

[NotebookLM @NotebookLM]

Because you wouldn’t let it slide… these are rolling out today for our most requested feature:

Prompt-Based Revisions: Tweak, tailor, and tune your slides just by prompting the revisions you want

PPTX Support: You can now export your Slide Decks (Google Slides coming next!)

[Alex Smagin @asmagin_]

@SawyerMerritt Imagine how hilarious would it be, if Tesla would publicly offer MKBHD to buy the first model for $29999 in the end of December 2026?

[Chrys Bader @chrysb]

welp, it was fun while it lasted folks

[Tech with Mak @techNmak]

Google just killed the document extraction industry.

LangExtract: Open-source. Free. Better than $50K enterprise tools.

What it does:
→ Extracts structured data from unstructured text
→ Maps EVERY entity to its exact source location
→ Handles 100+ page documents with high recall
→ Generates interactive HTML for verification
→ Works with Gemini, Ollama, local models

What it replaces:
→ Regex pattern matching
→ Custom NER pipelines
→ Expensive extraction APIs
→ Manual data entry

Define your task with a few examples.
Point it at any document.
Get structured, verifiable results.

No fine-tuning. No complex setup.

Clinical notes, legal docs, financial reports, same library.

This is what open-source from Google looks like.

[Itamar Golan 🤓 @ItakGol]

Disaster is coming.

Thousands of ClawdBots are live right now on VPSs… with open ports to the internet… and zero authentication.

This is going to get ugly.

If your agent can:
- browse the web
- call tools
- access files/secrets
- hit internal endpoints

…then an unauthenticated public endpoint is basically “please take over my bot”.

This isn’t theoretical. The internet is a nonstop scanner.

Fix it today:
1) close the port / firewall to VPN or IP allowlist
2) add auth (JWT/OAuth, at least a strong secret) + TLS
3) rotate keys (assume compromise)
4) rate limit + logs + alerts

Agents are powerful. Demo-grade deployments on the open internet are not.

[Robert Youssef @rryssf]

DeepMind just did the unthinkable.

They built an AI that doesn't need RAG and it has perfect memory of everything it's ever read.

It's called Recursive Language Models, and it might mark the death of traditional context windows forever.

Here's how it works (and why it matters way more than it sounds) ↓