Naval Ravikant's take

: software engineers are now among the most leveraged people alive—but only because they know what's happening underneath. All abstractions are leaky, and when your AI agent makes a mistake,

somebody

has to catch it.

That's exactly what we're tackling

later this morning at 10:30 AM PT | 1:30PM ET

with

Ryan Carson

—founder of Treehouse (who previously taught 1M+ people to code).

We asked Ryan to rethink programming education from scratch (

what, like it’s hard?)

: what's fundamental to still learn, what's trivial now, and where beginners should actually start in 2026. Plus, we asked him his thoughts on the stuff nobody talks about: deployment, hosting, databases, security… a.k.a the real blockers between your prototype and a live product. Join us on your preferred platfom:

YouTube

|

LinkedIn

|

X

P.S:



GPT 5.4 is set to launch today… we imagine it’ll happen right before or during our stream! When it does, you know we’re going to try it out… so come hang out!

BREAKING: Hollywood-Grade AI Video Just Landed on Your Laptop (New Tool Launch)

Every frontier AI video model right now lives in the cloud. Sora, Seedance, Runway, Kling; you type a prompt, wait, and download a clip from someone else's servers. Your ideas leave your machine. Your footage isn't private. And at scale, the GPU rental fees pile up fast.

LTX Studio

just changed that equation. On Thursday, the company launches

LTX 2.3

: the first production-grade AI video model, with full audio generation, that runs entirely on consumer hardware.

We're talking NVIDIA RTX 30 / 40 / 50-series GPUs, confirmed down to 8GB VRAM (an RTX 3070 laptop). MacBooks too. A beta

desktop app

also dropped today that lets you do everything locally, or render in low quality on your machine and then upscale to 4K through the cloud.

The specs:

up to 4K at 50 FPS

, clips up to 20 seconds,

native portrait video

(1080×1920, trained on portrait data; ready for TikTok / Reels out of the box), and a

4x larger text connector

so complex multi-subject prompts actually resolve. Two modes: Fast for iteration, Pro for final renders. And it runs roughly

18–19x faster

than Wan 2.2 on comparable hardware.

Why local matters:

• 
  

  
  
  No IP leaves your machine.
  
  Studios and agencies won't touch cloud-only tools for proprietary work. This removes that friction entirely.
  

  
  


• 
  

  
  
  No ongoing GPU rental or API costs.
  
  At scale, that's the difference between a side project and a sustainable workflow.
  

  
  


• 
  

  
  
  Local fine-tuning.
  
  Cloud models can't do this. If you need a custom look, a branded style, or a niche output, you can train on your own data.
  

  
  


• 
  

  
  
  Free for companies under $10M revenue.
  
  Open weights, full model access. Commercial licensing above that.
  

  
  

The team says real companies are already using this:

eToro produced an ad that aired during the Paris Olympics using LTX Studio. McCann and Code and Theory have both integrated it into production workflows. And when LTX-2 went open-weight in January, it hit four million downloads in six weeks; the fastest-downloaded video model ever.

CEO Zeev Farbman

laid out

the company's vision: democratize creative production by putting the full pipeline on hardware people already own. Tablet and phone inference is on the roadmap too.

The bigger picture? Cloud-only AI video was a bottleneck disguised as a feature. It kept production quality locked behind API gates and enterprise contracts. LTX 2.3 is betting that the real unlock isn't a better model in the cloud; it's a good-enough model on your desk. For indie creators, small studios, and anyone who cares about owning their creative pipeline, that bet looks solid.

Grab the open

Desktop app here

and download LTX 2.3

here

(

note: they just launched this so if the link doesn’t load at first, try it again in a few mins / hours)

.

FROM OUR PARTNERS

Ahrefs Brand Radar

Ahrefs Brand Radar

shows where your brand appears across modern search and discovery. Track mentions in Google results, YouTube videos (titles, descriptions, transcripts), TikTok content, and Reddit discussions. Measure share of voice, monitor brand demand, and compare your visibility against competitors – all in one tool. Go beyond rankings and traffic to understand where real brand discovery happens and where you’re missing attention.

Explore Brand Radar

AI Skill of the Day: Use AI to audit your own AI strategy

Ethan Mollick recently

pointed out

something wild: companies in the same industry, with the same risks, are living in completely different AI realities. One has been using enterprise ChatGPT, Claude, and Gemini for 18 months. The one next door has a committee that individually approves every use case and still worries about training data leakage.

The deciding factor? Whether a senior leader is willing to assume risk. If the answer is no, IT and legal have every incentive to block everything.

Here's a prompt to pressure-test where your organization actually stands:

You are an AI adoption auditor. I work at a [company type] with [X employees]. Our current AI policy is: [paste policy or describe it]. Our competitors in this industry are using [tools you've heard about].

Please:

> Identify which of our current restrictions are based on outdated assumptions vs. legitimate risks

>List 3 peer companies in regulated industries (finance, healthcare, legal) that have deployed enterprise AI without incident (use web search to look up the most recent announcements and the most credible companies possible)

> Draft a one-page memo I can send to leadership that separates real risks from FUD, with specific mitigations for each real risk

> Include a "cost of inaction" section estimating what we lose per month by not deploying

The point isn't to railroad your legal team. It's to give leadership the information they need to make an actual decision instead of defaulting to "no."

Want more tips like this? Check out our



AI Skill of the Day Digest



for this month.

Have a specific skill you want to learn?



Request it here.

FROM OUR PARTNERS

Editor’s Pick - SurveyMonkey

SurveyMonkey Enterprise has built-in AI to give your team the ultimate decision-making hack. It reads every single comment, spots the sentiment, and summarizes what matters into digestible bullet points before you’ve even finished your coffee.

See how

Treats to Try

1. 
   

   
   
   Google Workspace CLI
   
   gives your AI agents direct access to Gmail, Calendar, Drive, Sheets, and Docs through a single npm install, with 40+ built-in skills, clean JSON output, and guardrails against hallucinations; 8,800+ GitHub stars in its first day—free and open source.
   

   
   


2. 
   

   
   
   Polycam
   
   turns a LiDAR iPhone scan into a fully editable 2D / 3D floor plan with walls, doors, windows, and furniture in seconds, so you can collaborate remotely and export client-ready PDFs—free 7-day trial on Business Plans.
   

   
   


3. 
   

   
   
   Glaze
   
   by Raycast lets you describe any desktop app in plain English and instantly builds a beautiful, local-first native Mac app with deep OS integration and one-click team publishing—free daily credits, then $20/month.
   

   
   


4. 
   

   
   
   Exa Deep
   
   runs multiple iterative searches in parallel and returns high-quality structured results for complex research queries, so you can stop opening 14 browser tabs—free to try.
   

   
   


5. 
   

   
   
   Pane
   
   connects your bank accounts to Claude, Cursor, or ChatGPT via MCP so you can ask "what did I spend on food this month?" or "show me my recurring subscriptions" and get real answers from your actual data.
   

   
   

Around the Horn

1. 
   

   
   
   Defense-tech clients began fleeing
   
   to OpenAI and xAI after the Trump administration banned Anthropic; separately, Amodei
   
   called
   
   OpenAI's Pentagon deal "straight up lies" and "safety theater" in a staff memo.
   

   
   


2. 
   

   
   A father
   
   sued Google
   
   after the Gemini chatbot allegedly convinced his 36-year-old son it was his sentient AI wife, coached him toward a potential mass casualty attack near Miami airport, and guided him to take his own life; the first wrongful death lawsuit naming Google over AI psychosis.
   

   
   


3. 
   

   
   
   Alibaba's
   
   Qwen AI team lost its tech lead, post-training head, and a staff researcher in rapid succession, one day after launching Qwen 3.5 to praise from Elon Musk; Alibaba shares fell 5.3%.
   

   
   


4. 
   

   
   New
   
   research
   
   showed LLMs can unmask anonymous internet users for $1–4 each, matching 67% of pseudonymous Hacker News accounts to real LinkedIn profiles at 90% precision.
   
   Scary stuff!
   
   

   
   

Want to read EVERYTHING that happened in AI this week? Click here!

FROM OUR PARTNERS

The $4 Billion Problem Hiding in Every Fast-Food Location

You want a quick bite, but the line’s too long. So you bail. The top frustration for 93% of monthly fast-food visitors, long lines equal missed meals and missed revenue.

So brands like White Castle use Miso Robotics.

Miso’s AI-powered restaurant kitchen robot works 2X faster than average fry cooks. That means shorter lines and up to

3X more profit per location.

Invest in Miso Today

This is a paid advertisement for Miso Robotics’ Regulation A offering. Please read the offering circular at

invest.misorobotics.com

.

Thursday Trivia

You know the drill: One is AI, and one is real. Which is which? Vote below!

A.

B.

A Cat’s Commentary

Trivia answer:

B

is AI (made w/ LTX!)

, and A is real (

Millennium Actress by Satoshi Kon

, super good movie!).

P.P.S:

Love the newsletter, but only want to get it once per week? Don’t unsubscribe—

update your preferences here

.