maxwell

25| i talk about AI & Tech [email protected]

[Lightreel AI @lightreelai]

Introducing the largest Meta ads dataset, ever.

Over 3 million ads analyzed... yesterday

Ask Lightreel:

"Find me all Granola paid influencers"
"What do good street interviews do?"
"Write better hooks for my ads"

Try it for free!

(PS — Every pixel in this video is a Meta ad)

[maxwell @maxwell_lenoir]

@karlarboledas I follow him now

[maxwell @maxwell_lenoir]

@wairazzz Higgs is always improving 🤯

[máira @wairazzz]

I used GPT-6 Astra + Higgsfield to recreate Twilight with calendar events.

My schedule finally has the right amount of drama.

[Mark Ajzenstadt @mardehaym]

A PE-backed company that builds the kitchen-operations software national restaurant chains run on came to us with a prototype a non-engineer had built with AI.

It couldn't handle production load. Six months later, one of our engineers had it live and stable.

"Just use AI" is not the lesson here. They already used AI. That's exactly how they got a prototype that couldn't ship. I want to walk through the part that actually mattered.

The platform is the operational backbone a location runs on: demand forecasting off sales history, prep plans, inventory and expiry tracking, timers and tasks, all of it on the tablets the crew works from.

Their customers are national quick-service chains, the household names you'd recognize on any corner. And a contract with one of the largest restaurant groups in the world was on the table, a rollout that could reach hundreds of locations.

The prototype was the only thing standing between them and it.

That prototype was code an AI had written for someone who couldn't review it. It looked like a product. It wasn't a system. Unstable, impossible to extend, nowhere near able to hold up with hundreds of locations running it at once.

We put one AI Velocity Pod on it. One engineer and a delivery manager. Not a team of ten.

We didn't throw the prototype away and we didn't worship it. We kept what the AI got right, rebuilt everything it got wrong, in production, behind review gates, in a daily joint loop with their side.

Every change read by a human who owned what happened when a restaurant ran it live.

Their non-engineer builder stayed on the project.

Instead of sidelining him, our engineer taught him to work inside the system properly, Git, reviews, database migrations, so the changes he shipped with AI stopped breaking the ones we did.

The kind of build that used to take a team of four a couple of years, with no guarantee at the end of it. One pod did it in six months.

It's in production now and being demoed to that restaurant group as they weigh the rollout.

And the part that matters most to me: this client worked with us years ago, left, and came back for the hard version.

AI gets a non-engineer to a prototype. It does not get a company to a platform.

A prototype proves the idea. A system survives the scale. We build the second one.

[maxwell @maxwell_lenoir]

@S0N_IA esto no es ciencia ficción, es un empleado renunciando por lo que vio y da miedo

[karla @karlarboledas]

Este chico explicó cómo creó 3 webs que le generan 100,000$ al mes.

En 25 minutos muestra cómo lo hizo, paso por paso. Totalmente gratis.

Idea → Web → Monetizar

guárdalo y sígueme para más → @karlarboledas

no lo dejes pasar y aprende de el.

[maxwell @maxwell_lenoir]

Unreliable AI agents disrupt live workflows and damage customer trust.

State Machines simulates full Salesforce operations, giving teams complete confidence that their agents will accurately update records and close deals before going live.

[gus @igus_ai]

SEGUID A ESTA PERSONA, NO DIGO NADA MÁS…

[maxwell @maxwell_lenoir]

@vinayjain404 41 people complaining about the same thing isn't a complaint, it's a positioning brief with a sample size

[Cresta @cresta]

We’re welcoming Super Bowl champion, Bay Area legend, and media icon @MoneyLynch to Team Cresta.

He earned the name Beast Mode by being unstoppable. Now your customer experience can be too.

Call the right play. Solve the problem. Never fumble.

Put your customer experience in Beast Mode with Cresta.

#CXinBeastMode

[maxwell @maxwell_lenoir]

Unreliable AI agents disrupt live workflows and damage customer trust.

State Machines simulates full Salesforce operations, giving teams complete confidence that their agents will accurately update records and close deals before going live.

[state machines @statemachineshq]

Introducing State Machines.

The first infrastructure to spin up enterprise environments.

Any enterprise app, recreated for agents. Run thousands of environments in parallel, each with its own state.

https://t.co/ApBSmZRe1h

[state machines @statemachineshq]

Introducing State Machines.

The first infrastructure to spin up enterprise environments.

Any enterprise app, recreated for agents. Run thousands of environments in parallel, each with its own state.

https://t.co/ApBSmZRe1h

[Antioch @antiochrobotics]

Today, we’re announcing Antioch’s $32 million Series A, led by @GreylockVC with participation from @A_StarVC, @Category_VC, @BoxGroup, @IcehouseVenture, and angels.

While AI has drastically accelerated software development, physical autonomy has been constrained by slow, expensive hardware-based development. Antioch enables physical AI teams to build, test, and validate systems at the speed of software.

We’re proud to be working with leading teams including @amazon @ring, @NVIDIARobotics, and @nebiusai as we make scaled, high-fidelity simulation the standard for physical AI development.

Our sincere thanks to the legion of investors, customers, partners, and advisors who are making this step change possible.

We’re scaling rapidly and hiring across simulation, infrastructure, machine learning, 3D graphics, go-to-market, and operations.

[maxwell @maxwell_lenoir]

Este tipo te enseña cómo monetizar un canal de YouTube con IA sin mostrar tu cara.

Tan sólo 17 minutos paso a paso.

Guárdalo, te será útil.🔖

[maxwell @maxwell_lenoir]

Este tipo te enseña cómo monetizar un canal de YouTube con IA sin mostrar tu cara.

Tan sólo 17 minutos paso a paso.

Guárdalo, te será útil.🔖

[Alisher S @gpumaxxer]

we reached AGI. Higgsfield is using GPT-6 Astra to make N3on stream forever.

we’re testing a new feature on @higgsfield_ai: endless AI livestreams.

new video gets generated as you watch. there’s always a next moment.

AGI is here and its first job is being N3on.

[Axultan @axultan]

GPT-6 Astra is now powering @N3onOnYT’s infinite livestream with Higgsfield.

This is the first-ever infinite, interactive, highly realistic, actually watchable, real-time AI livestream.

30K+ viewers on Kick and 500K+ impressions on X in under an hour.

AGI is here.

[maxwell @maxwell_lenoir]

@karlarboledas Gracias karla🙂‍↕️

[karla @karlarboledas]

🚨NVIDIA está dando GRATIS 5 de los modelos chinos de IA más avanzados

Sin tarjeta.
Sin suscripciones.
Sin pagar por cada modelo.

Una sola API key te da acceso a los 5.

Esto es lo que puedes usar por $0:

DeepSeek V4 Flash → inferencia ultrarrápida
MiniMax M3 → coding y desarrollo
Qwen3.5-397B → razonamiento avanzado
Kimi K2.6 → agentes y workflows complejos
GLM 5.1 → modelo general para el día a día
Pero lo realmente interesante es cómo puedes utilizarlos.

Los 5 funcionan con una API compatible con OpenAI, así que puedes conectarlos a tus herramientas y workflows actuales sin tener que cambiar toda tu infraestructura.

Cómo empezar:

1. Ve a https://t.co/O7ieVinJe3
2. Crea y verifica tu cuenta
3. Genera tu NVAPI key
4. Configura https://t.co/2DaAocJucT como base URL
5. Elige cualquier modelo y empieza a construir

Los modelos disponibles:
• minimax-m3
• qwen3.5
• kimi-k2.6
• glm-5.1
• deepseek-v4-flash

Y hay algo todavía más interesante:

La misma API key te da acceso a más de 100 modelos del catálogo de NVIDIA.

Puedes usar:

→ DeepSeek para velocidad
→ Qwen para razonamiento complejo
→ Kimi para agentes
→ MiniMax para coding

Todo desde una única API.
5 modelos chinos de IA de primer nivel.
Una sola API key.
$0 para empezar.

guárdalo y sígueme para más → @karlarboledas