Lorenzo Toscano

🤖 AI Pioneer | Entrepreneur | Author | Building Agentic AI | 30+ yrs in code & innovation | DM for partnerships

[Kirk Borne @KirkDBorne]

[Download 585-page PDF eBook]

Game Theory: https://t.co/8NySL3GFmu
—————
#GameTheory #Gamification #Mathematics #Statistics #Probability

[Turing Post @TheTuringPost]

6 guides to understand how LLMs actually work

- What is a Token?
- How token taxonomy affects your bill
- Embeddings (including RoPE)
- Agentic Vector Databases
- Attention: Mechanism, QKV, and KV Cache
- From tokens to answers: What actually happens during LLM inference

Links ↓

[Mark Kretschmann @mark_k]

Clearly, Google Omni has been wildly underrated.

Here it's turning a normal human hand into a live anatomy demo! Letting you see the muscles, tendons, and cartilage as if the skin were removed.

Brilliant for educational purposes!

[Claude @claudeai]

Each agent ships with the connectors, skills, and subagents the task needs, ready to use as-is or adapt to your firm's own standards.

Read more: https://t.co/DKPBIaWlbk

[Kelly Boesch🏳️‍🌈 @kellyeld]

I wrote this song after hearing another woman tell me her idea was stolen by a man in a mtg. Great article about this in #forbes. I used an abstract idea for the video. Additional production by Marshall Altman. #ai #aiart #music #aianimation

[Bülent DMK @im_bulent]

C’est peut être le début le plus lowcost possible de la robotique personnelle

[Nico @nicos_ai]

Han clonado Claude Design y puedes usarlo gratis y sin limites.

Se llama Open Design, un proyecto open source que te deja usar Claude para workflows de diseño sin pagar.

Sin suscripciones.
Sin límites (como la versión oficial).
Acceso total.

Esto es lo que puedes hacer:

— Generar diseños UI/UX con Claude
— Convertir prompts en diseños reales
— Sustituir herramientas de diseño caras en muchos casos
— Personalizarlo completamente (es open source)

Está hecho para devs, indie hackers y creadores que no quieren quemar dinero en herramientas.

De esos repos que pasan desapercibidos hasta que de repente todo el mundo los usa.

Si usas AI + diseño, tienes que probarlo

Enlace abajo 👇

(guárdalo antes de que explote)

[Lorenzo Toscano @BEmatic]

https://t.co/zAk4TUaGj7

[Lorenzo Toscano @BEmatic]

https://t.co/zAk4TUaGj7

[Sawyer Merritt @SawyerMerritt]

Watch how perfectly Tesla FSD (Supervised) handles all these pedestrians and bikers on narrow roads in Amsterdam, the biking capital of the world. It's a thing of beauty. Lot of fun to watch all these FSD videos coming from the Netherlands.

[leopardracer @leopardracer]

https://t.co/nZCe073239

[DAIR.AI @dair_ai]

https://t.co/VbnHQG4Mh8

[Paul Couvert @itsPaulAi]

You can now fine-tune Gemma 4 (and 500 other open source models) in a free Google Colab 🔥

1. Open the Colab notebook below
2. Run the blocks to launch Unsloth Studio
3. Choose a model and dataset
4. Hit 'Start Training'

And you're done!

[Kevin Gu @kevingu]

as agents pass 99th percentile human performance, our intuitions about harness design become the wrong prior. like AlphaZero, they should discover from first principles. AutoAgent is open source:
https://t.co/I7ajRbatg6

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[Matt Dancho (Business Science) @mdancho84]

RIP document extractors.

Google just released LangExtract: Open-source. Free. Better than $100K enterprise tools.

Here’s what it does: 🧵

[Muhammad Ayan @socialwithaayan]

🚨 BREAKING: Qwen3.5-Omni just dropped a mind-blowing emergent ability:

Audio-Visual Vibe Coding
No specific training.
Just raw power.

→ Turn whiteboard brainstorming videos into fully functional websites
→ Turn gameplay screen recordings into playable code

Vibe Coding just went from "cool idea" to zero-barrier reality.

You sketch it. The AI builds it. Instantly.

The future of coding is here and it's wild.

But wait… there's way more 👇

[Lorenzo Toscano @BEmatic]

@GithubProjects A2UI focuses on agents describing UIs.

Another interesting direction is PLASMA → https://t.co/I23vESwEYP

Here the UI is event-sourced and evolves through interaction, so the application grows as a history of agent actions rather than a static JSON tree.