olivier peypin

Entrepreneur , chercheur en verite

[Black Bond PTV @BlackBondPtv]

🚨 LE SCANDALE DU SIÈCLE : TOUT LE MONDE N’A PAS REÇU LE MÊME VACCIN

Le Parlement Européen crache le morceau :

Le vaccin COVID n’était pas le même pour tout le monde.
Certains lots étaient légers.
D’autres, selon les lanceurs d’alerte, chargés à mort.

Deux fichiers d’effets secondaires.
Un pour le public.
Un autre gardé par les labos.

Et l’EMA le savait déjà :
ces vaccins n’ont jamais été autorisés pour arrêter la transmission.

Le discours, c’était « tu le fais pour les autres ».
La réalité, c’était autre chose.

Le crime du siècle.

Crédit : @WiretapMediaCa

[Miles Deutscher @milesdeutscher]

Send this crypto research prompt to Grok Bot.

It will probably make you a lot of money.

You're welcome:

[Adam Girard (Gilet Jaune) @AdamGirard1990]

L'Etna a craché en 3 jours plus de CO2 que 10 ans de sacrifices imposés par l'UE. Tout leur Green Deal anéanti par un volcan!

Mais c'est TA voiture qui pollue! Pourquoi? Parce que l'arnaque écolo leur permet de te racketter à vie avec ZFE, taxes et malus.

[Antonio Rodríguez @Antonio_RodriIA]

🚨 ALGUIEN ACABA DE CARGARSE LA INDUSTRIA INMOBILIARIA

Un tipo escaneó una casa entera con su teléfono. La subió.

Ahora cualquier persona del mundo puede recorrerla desde una pestaña del navegador. Sin app. Sin realidad virtual. Sin agente. Sin cita.

Clic → estás dentro. Cada habitación. Cada ángulo. Cada sombra. Fotorrealista.

Las cifras son una locura:

* Comisión de un agente por una casa de 500.000 $: 15.000 $
* Coste de hacer este escaneo: ~200 $
* Tiempo para “visitar” 50 casas: una tarde
* Tamaño del archivo: menor que un TikTok

La tecnología también es increíble:

Se llama 3D Gaussian Splatting. En lugar de polígonos —como suelen renderizar los videojuegos— utiliza millones de pequeños “splats” luminosos con información de color y profundidad.

La IA reconstruye la realidad a partir de tus fotos. El resultado se carga en un teléfono y parece que estás ALLÍ.

Y la oportunidad de negocio es todavía más interesante:

Los freelancers ya están cobrando entre 300 y 800 $ por escaneo para inmobiliarias, Airbnbs, espacios para eventos, concesionarios de coches y museos.

Una persona + un teléfono + un fin de semana = un negocio.

100 % open source. Construido sobre PlayCanvas.

[boubacar36 @boubakar58577]

🚨🔴🔥‼️
2 MOIS.....
DEUX MOIS QUE LES PIRATES SE BALADAIENT DANS LES SERVEURS DES IMPÔTS
700 000 foyers.
Revenus fiscaux, taux d’imposition, composition familiale, adresses… Tout dans la nature.
Revendu sur le dark web.

😡Et pendant ce temps ? Silence radio à Bercy. Zéro communication. Zéro alerte.
C’est seulement quand ZeroBytes a revendiqué le coup et que les réseaux ont explosé que la Macronie a soudain découvert l’urgence.
Excuses du ministre. Cellule de crise. Mails aux victimes.

😳Sans 𝕏 sans la pression des Français, on aurait encore eu droit au grand silence administratif.
Voilà le vrai visage de l’État : incapable de protéger les données de ceux qui le financent… mais très fort pour taxer.

[AlertesNewsFR @AlertesNewsFR24]

🤯 « LE VOL DU SIÈCLE »

Julian Muller (ancien inspecteur des impôts) s’adresse aux parents :

« Il y a 9 000 milliards d’euros qui vont se transmettre. L’État peut en prendre jusqu’à 45 % juste avec les impôts sur l’héritage. »

« Tout ce que vous avez mis de côté pendant 40 ou 50 ans de travail va servir à rembourser la dette de l’État. »

Et ça commence dès 115 000 €.

La solution est simple : allez voir un notaire et faites une donation.

Il explique tout 👇

[RetroChainer @RetroChainer]

10 GITHUB REPOS TURN CLAUDE + OBSIDIAN INTO A SECOND BRAIN. THE HONEST TAKE: YOU ONLY NEED 2 OF THEM.

someone catalogued 10 free repos that give claude a real memory with obsidian as storage. install five and you lose a weekend to config debt. here's the cut.

every repo on that list is one of three ideas:

the karpathy wiki: raw files in, claude compiles them into linked pages once and never re-reads the source, cutting token spend 70-90% on repeat lookups

skills: drop a SKILL.md, claude auto-loads the right instructions only when your ask triggers it

the mcp bridge: a live connection so claude reads, or writes, your vault in real time

now the honest part. you need two, not ten.

one compiler from the karpathy family to turn raw sources into structured knowledge - AgriciDaniel/claude-obsidian is the cleanest, ekadetov/llm-wiki the most minimal.

one mcp bridge to give claude runtime access across every tool - read-only if your notes are precious, read-write if you want the vault to build itself while you sleep.

that's the whole system. a compiler plus a bridge.

the line that ends the debate: raw is source code, the wiki is the compiled product. rag without that compilation is just grep with vibes.

no five overlapping installs, no config debt, no rag pulling answers out of your mess.

the uncomfortable part isn't that there are 10 repos. it's that 8 of them are variations of 2 ideas, and the win was always compile first, retrieve second.

follow @RetroChainer for the two-tool cut instead of the ten-tool rabbit hole, and bookmark this before you install your third.

[MIKE @mikenevermiss]

https://t.co/Y69Pcyy4Rr

[Mahax @Mahaximus_]

Google just dropped a 1-hour course on Graph Engineering: from agents to Loops to full automation

00:00 – Build your first AI agent
08:24 – Build agent memory
28:34 – Agentic loops
40:04 – How to build MCP
1:00:22 – Graph engineering

This 1-hour watch replaces any $500 course you could pay for.

Watch it, then try your first graph with the step-by-step guide below

[Lunar @LunarResearcher]

Google just released a free 2-hour course on full Graph Engineering.

How to go from one prompt to 100 agents running inside one graph:

17:44 - Build your first AI agent
39:30 - Run agents with loop engineering
1:12:38 - Turn agent loops into graphs
1:34:26 - Build agents that throttle themselves
1:55:05 - Orchestrate the full multi-agent system

Most people build one agent and stop there.

Google is teaching the full stack:

Prompt → Agents → Loops → Graphs → Multi-Agent Systems

Single agents are the old workflow.

Self-regulating agent graphs are the new one.

This 2-hour watch is worth more than most paid agent engineering courses.

Bookmark and watch it today

Then read the full architecture below ↓

[Khairallah AL-Awady @eng_khairallah1]

https://t.co/X5BAxRdTKh

[Noisy @noisyb0y1]

ANTHROPIC MAPPED OUT 8 LEVELS OF AI AGENTS - AND ANYONE CAN REACH LEVEL 8 WHERE THE SYSTEM EARNS $70K WITHOUT YOU

anyone can reach level 8 - one solo developer with the right architecture gets there in a few weeks

level 1 - the agent can do one action - and 80% of developers call this a product

level 3 - 50 tools in context at once - the model gets confused - most teams get stuck here and lose $30K/month

level 4 - Opus 5 grades its own output and rejects before you see it - the first level where quality stops depending on luck

level 5 - Opus 5 plans, Fable 5 executes in parallel - one brain, many hands

level 7 - while you slept the agent did $3,000 worth of work and sent the report before you woke up

level 8 - the fleet finds its own mistakes, patches the graph and tomorrow it is smarter than today

bookmark and read below - 8 steps from level 1 to $70K/month without you

[Avid @Av1dlive]

this is f**king insane

a free github repo by Jack Dorsey (Co-Founder of Twitter) with 14.4K stars just dropped the entire "ai-agent" framework for running businesses

here is how you set it up:

1.clone the repo
2. self-host the server : channels, search, git, automation all live there
3.add your agent to a channel like a teammate, scope its key, let the team steer it live

save and bookmark this no matter what

[Avi Chawla @_avichawla]

RAG vs. Graph RAG, explained visually!

RAG has many issues.

For instance, imagine you want to summarize a biography, and each chapter of the document covers a specific accomplishment of a person (P).

This is difficult with naive RAG since it only retrieves the top-k relevant chunks, but this task needs the full context.

Graph RAG solves this.

The following visual depicts how it differs from naive RAG.

The core idea is to:
- Create a graph (entities & relationships) from documents.
- Traverse the graph during retrieval to fetch context.
- Pass the context to the LLM to get a response.

Let's see how Graph RAG solves the above problem.

First, a system (typically an LLM) will create a graph from documents.

This graph will have a subgraph for the person (P) where each accomplishment is one hop away from the entity node of P.

During summarization, the system can do a graph traversal to fetch all the relevant context related to P's accomplishments.

The entire context will help the LLM produce a complete answer, while naive RAG won't.

Graph RAG systems are also better than naive RAG systems because LLMs are inherently adept at reasoning with structured data.

Once the right architecture is in place, the next leverage point is efficiency.

Most RAG architectures rely heavily on vector search, and that layer can be made 32x more memory efficient using binary quantization.

I covered the full implementation in the article below.

👉 Over to you: Which RAG architecture are you running in production?

[codila @0xCodila]

This free 6-hour course gets you Anthropic official "Claude Certified Architect" certificate

with it you'll be hired at $250K/year in Anthropic's ecosystem

10:17 - set up your first Claude SDK in 10 min
47:36 - agentic loop foundations
1:52:53 - graph engineering
2:50:00 - tool use - make your agent act, not just think
4:15:46 - advanced agent patterns

after watching the full freeCodeCamp tutorial (12 hours) - go to Anthropic's website and take the exam - you're ready

watch it today - then read how to become a knowledge graph architect in the article below ↓

[angel @angeldot_]

UN EQUIPO ACABA DE DESPLEGAR 15 AGENTES AUTÓNOMOS EN LOOP A PARTIR DE UN SOLO PROMPT USANDO GRAPH ENGINEERING

La mayoría de desarrolladores todavía crean sistemas multiagente a mano, escribiendo lógica independiente para cada tarea.

Con Graph Engineering, eso cambia por completo.

Un graph central define la arquitectura y despliega los 15 agentes simultáneamente.

Un único prompt de apenas 200 palabras genera toda la estructura y enruta 120 conexiones entre agentes de forma automática.

En lugar de romperse cuando reciben instrucciones contradictorias, los agentes se corrigen entre sí gracias a un estado compartido que se actualiza continuamente.

Gestionar una malla de 15 nodos exige un alto rendimiento y una API con baja latencia, ya que todos los agentes intercambian contexto en tiempo real.

Mira el vídeo y luego aprende como crear tus propios grafos con el artículo de abajo

Así es como funciona esta arquitectura multiagente basada en grafos y automatizada en tiempo real ↓

[Semichenko @semichenkko]

THIS IS WHAT HAPPENS WHEN YOU CONNECT JARVIS TO A LIVING NEURAL NETWORK.

Claude Code wasn’t enough.

First we built an entire neural network for your computer because Obsidian feels dead.

Then we connected it to Jarvis.

Now you can just say:

“Hey Jarvis, explain to me what I’m wearing right now… and once you’re done, tell me how Neuralink actually works.”

And it instantly reads your entire brain folder (.md files: notes, tasks, personality, projects, history), understands context, and answers like a real teammate.

This isn’t storage.

This is a persistent, thinking memory system that fires neurons in real time and never forgets who you are.

From static notes → to an active second brain that lives with you.

[Gipp 🦅 @gippp69]

THIS 38,000-STAR GITHUB REPO TURNS ONE AI AGENT INTO A REAL TEAM THAT CAN BRANCH, VERIFY ITS WORK AND WAIT FOR YOUR APPROVAL

most people still run agents as one long chain where every step waits, one failure kills the run and the full workflow starts again

Task → Planner → 5 Researchers in Parallel → Skeptic → Writer → Human Gate

LangGraph gives every node one job while a shared state carries the findings, decisions and context through the entire system

the skeptic can reject an unsupported finding and route the work back before it contaminates the final report, while independent branches keep moving

if the run crashes, durable execution resumes from the saved state instead of rebuilding everything, then human-in-the-loop pauses the graph before anything expensive gets sent or published

bookmark this repo and watch one prompt turn into an actual org chart for AI agents

[Swati Gupta @hrswatigupta]

ANTHROPIC ENGINEER:

"DON'T PROMPT CLAUDE. BUILD A SYSTEM THAT PROMPTS ITSELF."

In this free workshop, he explains why most people use Claude the wrong way.

And how to turn one AI into an entire team of AI agents.

He covers:

• Proper CLAUDE.md setup
• Plugins almost nobody uses
• Advanced prompt caching (95% cache hit rate)
• Why starting every chat from scratch is a mistake
Worth more than most paid Claude courses.
Watch this and bookmark it.

[SilenceÇaPrompt @SilenceCaPrompt]

Ce mec te montre comment monter ton app de A à Z avec Claude en 8 minutes