Open This Before Jan 2
The episode highlights the rare, distraction‑free period between Christmas and New Year’s as an ideal time to decide whether to ship a real AI product in the coming year. It outlines a two‑part learning path—a free 10‑hour LLM Primer and a full‑stack AI Engineering course—designed to give listeners practical judgment on prompting, retrieval‑augmented generation, evaluation, and agent use, then guide them to build and deploy a production‑grade LLM pipeline. The host emphasizes that completing this bundle by Jan 2 will turn holiday resolutions into tangible proof of capability, positioning listeners to start 2026 as effective AI engineers rather than just theorists.
Showcase AI Work Publicly; Jobs Find You
I didn’t get my first AI job by applying anywhere. It started with YouTube. 🎥👇 I was posting simple research explainers on YouTube when one day, the founder of a startup left a comment: “Can we talk?” He noticed I was also from Québec...
First University Talk: Turning AI Work Into Teach‑Ready Insight
I gave my first university talk this week at the University of San Diego. I was genuinely stressed. Not because the content was hard, but because it was the first time I had to turn what we do at Towards AI into...
Speech to Text Is Harder Than You Think
The video tackles a misconception that speech‑to‑text (STT) is merely a matter of converting audio into words. It argues that for production voice agents, transcription is only the first step; the real battle lies in extracting precise entities, handling latency,...
Voice Agents Require Far More than Simple Speech-to-Text
Most people think speech to text is just turn audio into words. Anyone who has built a real voice agent knows... that's the easy part. https://t.co/vNSn0Tum1y
Write to Learn: Content Accelerates Mastery, Not Virality
Sharing content online is one of the highest-ROI habits you can build, and it has nothing to do with going viral. When I started, my goal wasn’t audience or money. It was learning. Content forces clarity. Saying “I want to learn...
The Hidden Skill Boost Behind Posting Online
The video explores the often‑overlooked benefit of publishing content online: it serves as a powerful learning accelerator. The creator explains that his initial foray into content creation wasn’t driven by audience size, revenue, or virality, but by a desire to...
AI Amplifies What You Already Have, Good or Bad
I’ve been watching a pattern lately: a lot of businesses are trying to “add AI” hoping it will magically fix everything. But here’s the honest truth: AI won’t save a bad business. It will simply reveal what’s already broken. If your operations are messy,...
Avoid Prompt Debt: Understand AI‑Generated Code, Don’t Outsource Thinking
Prompt debt might be the most 2025 kind of technical debt. It’s what happens when AI writes your code… but you never build the mental model behind it. Shaw Talebi calls it out clearly: LLMs can generate code, architecture, even some...
Ship Fast: Use Your Known Stack, Not New Tools
Stuck choosing a tech stack for your next AI project? You might be overthinking it. My friend Shaw Talebi is an AI engineer who ships a lot of small AI SaaS projects fast. His rule is simple: build with what you...
Decode Papers Quickly: Abstract, Figure, AI Summaries
I used to be terrified of reading research papers... until I learned this 👇 The first time I opened one, I thought: “There’s no way I can read this.” Too formal, too technical, too many acronyms - and English wasn’t even my first language. But...
ChatGPT’s Three Modes Adjust Reasoning Depth, Not Models
Do you know how Instant vs Thinking vs Auto mode works in ChatGPT? And what the actual difference is?👇 These aren’t different models. They are different reasoning modes for the same model. Here’s what actually changes when you switch modes: 1. Instant Mode (Fast): Fast mode...
AI-Generated Replies Make Twitter Conversations Feel Generic
Yeah, I’ve been seeing more replies that read like polished, template-style LLM outputs—overly neutral tone, generic phrasing, and lots of “great question” energy. It definitely changes how discussions feel. Next, if you want, I can generate you a short checklist...
AI Generates New Images via Compressed Latent Space
How does AI create images or ideas that never existed? Not by remembering everything. AI reduces the data into a small, meaningful code a compressed space where similar things stay close and different things spread apart. That space is called the latent space. The carousel below...
Chunk Retrieve Augment: RAG Cuts Hall
Here’s a simple breakdown of how a basic RAG pipeline actually works.👇 You start by breaking long documents into smaller, focused chunks and converting each chunk into an embedding vector. These embeddings capture semantic meaning which lets the system understand what each piece...
Build In-Demand Skills, Turn Entrepreneurship Into Low-Risk Experiment
Taking the leap into entrepreneurship doesn’t have to be reckless. Shaw Talebi’s story is the blueprint for doing it safely. He gave himself oxygen: 12 months of runway. That alone turned a “risky jump” into a reversible experiment. Worst-case scenario? He...
How To Try Entrepreneurship Without Ruining Your Life
The video tackles a practical question many aspiring founders face: how to dip a toe into entrepreneurship without jeopardizing financial stability. Using the experience of Shah Talibi as a case study, the presenter outlines a step‑by‑step framework that hinges on...
Master Python Projects with a Simple Four-Step Workflow
If you’re learning Python and building projects follow this simple workflow: Plan → Write → Test → Debug (+ Code with AI) This carousel breaks down the workflow beginners should use to build projects. If you want to learn Python + AI through hands-on...
Why Developers Should Not Ignore Kimi’s CLI
The video announces Kimi’s newest offering – a command‑line interface (CLI) agent that brings AI‑driven coding assistance directly into the developer’s terminal. Positioned as a competitor to established tools like Cloud Code, Gemini and OpenAI’s offerings, the Kimi CLI aims...
The Truth About Working for Yourself in AI
The video explores the realities of transitioning from a traditional AI role within a large corporation to running an independent AI consultancy, using Shah Talebbi’s journey from a data‑scientist at Toyota to founder of an AI education community as a...
UK Court Rules AI Training Not Copyright Infringement
Getty Images sued Stable Diffusion (Stability AI) in the UK and many expected this case to finally answer a big question: "Is training AI on copyrighted data illegal?" But the outcome surprised everyone. Getty claimed millions of their photos were used during training and...
Two AI Agents Streamline Product Shipping in Minutes
Everyone asks how to ship faster as an AI engineer. @ShawhinT nailed the answer. He’s a former senior data scientist turned entrepreneur and one of the most efficient AI builders I know — the guy is literally shipping full products,...
How Fast Can You Build With AI?
The video explores a streamlined workflow for AI engineers aiming to ship products at maximum speed, featuring Shah Terebi’s personal methodology. Terebi, a former senior data scientist turned AI educator, outlines how he leverages a combination of voice‑driven ChatGPT sessions,...
Gemini 3 & Nano‑Banana
Proudly repping that Gemini merch. Thanks for sending that @googleaidevs @GoogleAI ! Gemini 3 and all suites of models including nano banana 2.5 are a clear step forward and we use it in most of our projects and courses. Honestly, amazing progress...
Open-Weight Models Accelerate AI; China Poised to Lead
We keep talking about “open-source models" But honestly, that’s not where most of the real momentum is right now. What’s actually shaping the ecosystem today are "open models" Especially the ones released as open weights. Not fully open-source. Not fully closed either. Just open enough...
Upgrade Skills, Not Just Hardware, for AI Success
Everyone’s out here upgrading their setups today. New laptops, new GPUs, new toys. But the truth is simple: the real leverage isn’t the machine. It’s whether you actually know how to build and deploy AI on it. So if you’re upgrading...
Build Real AI Apps with 40% Off Today
If you’ve been wanting to jump from using AI tools to actually building real AI applications, this is your moment. The Black Friday window just opened for the Towards AI Academy, and it’s the lowest price we offer all year....
Community BFCM: $349 → $209 for Full-Stack AI Engineering
In this brief Black Friday announcement, Louis‑François Bouchard promotes a 40% discount on all AI engineering courses, highlighting the Full‑Stack AI Engineering program dropping from $349 to $209 as the flagship offer. He outlines how the course equips learners with...
KV Cache Speeds up AI After First Prompt
Your first question to an AI model takes a moment… But the next ones appear almost instantly, there’s a simple reason behind it. The model keeps a small snapshot of the work it already did. This is called the "KV Cache". When an AI...
AI Uses Embeddings, Not Memory, to Understand Context
Ever wondered how AI “remembers” your question… without having memory? 🤔 Every time you chat with an LLM, it somehow knows what you said before. But here’s the secret: It doesn’t remember your words. It understands meaning through something called embeddings. Embeddings are how machines...
LLM Production Guide Now Available in Simplified Chinese
Another cool milestone to share: My book Building LLMs for Production just got translated into Simplified Chinese!! …and I (again) can’t really proofread it. 😅 Still, it feels incredible. Posts & Telecom Press reached out last year asking to translate the book, and...
Essential ML Primer: From Basics to Cutting‑Edge Techniques
Since my review of the book actually made it inside, I'll just share it here: This book is an excellent starting point for beginners looking to understand the essential history and foundational concepts of machine learning. With well-structured code sections...
Is Synthetic Data Ruining LLMs?
The video centers on the contentious role of synthetic data in training large language models (LLMs) and vision‑language models (VLMs), featuring Leticia, a newly minted PhD who specializes in these areas. She weighs the benefits and drawbacks of generating artificial...
Synthetic Data Boosts Small Models, VLMs Need It
Synthetic data might be the most misunderstood topic in AI right now. Is it a cheat code for training better models or a trap that slowly collapses model diversity? Here's what @AICoffeeBreak, one of the sharpest minds in VLMs and...
VLMs Hallucinate when Text Dominates over Visual Grounding
What if your vision language model isn’t actually seeing… but mostly guessing from text? 👀 @AICoffeeBreak explains it perfectly: when VLMs rely too heavily on text, they start hallucinating answers based on the most common phrasing in their training data instead...
VLMs Rely Too Much on Text !!
The video highlights a growing concern in the field of vision‑language models (VLMs): they tend to lean heavily on textual cues at the expense of visual grounding, leading to what researchers call "text‑driven hallucinations." Leticia, a recent PhD graduate specializing...
Stay Relevant: Master AI Faster Than a Degree
𝗬𝗼𝘂 𝘄𝗼𝗻’𝘁 𝗯𝘂𝗶𝗹𝗱 𝗮 𝗰𝗮𝗿𝗲𝗲𝗿 𝗶𝗻 𝗔𝗜 𝗶𝗳 𝘆𝗼𝘂 𝗸𝗲𝗲𝗽 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗹𝗶𝗸𝗲 𝗶𝘁’𝘀 𝟮𝟬𝟮𝟬. A lot of people ask me if it’s still worth doing a Master’s to learn 𝘼𝙄. Honestly, for most people, it’s not about the degree anymore. It’s about staying...
AI Success No Longer Requires Academic Publications
Many people still believe you need to publish research to succeed in AI. That might’ve been true a few years ago, but things have changed 👇 I’ve done research during my Master’s and PhD. Most of my work never made it to...
Why AI “Forgets” Your Conversation
The video explains why large language models (LLMs) like ChatGPT appear to “forget” earlier parts of a conversation: they simply lack a true memory and are constrained by a fixed context window of only a few thousand tokens. When a...
Real-World AI Lessons for Production LLMs Inside
I packed years of real-world AI lessons into one book. If you want to actually build LLM systems that work in production, it’s all inside. Building LLMs for Production is available on Amazon, O'Reilly, and as an e-book (cheaper) on the...
Everything I Learned About LLMs in One Book
Louis‑François Bouchard, CTO and co‑founder of 2RD AI, introduces his new book *Building LLMs for Production*, a practical guide for developers who want to move from curiosity about large language models to building real‑world, value‑adding applications. The video outlines the book’s...
LLMs Sound Confident, Not Truthful—Question Their Answers
LLMs don’t know what’s true. They just try to sound helpful. If you ask the wrong question, you can still get a very confident answer. And confidence is not accuracy. Use AI to challenge your reasoning, not to replace it. But you need to be...
OK Computer Just Fixed My Slide Deck... By Itself
Today’s video spotlights Moonshot AI’s Kimi platform and its newly launched OK Computer agent mode, a free‑to‑use alternative to the market’s dominant chatbots. OK Computer transforms the traditional LLM from a token‑spitting text generator into an autonomous agent that...
Tavus PALs: AI Avatars with Multimodal Memory
The line between human and machine interaction keeps blurring, and Tavus’s new PALs launch pushes that frontier to the next level. What stands out isn’t just the realism with a nice avatar talking to us, it’s the engineering behind it: multimodal...
Kimi K2 vs GPT-5: The New DeepSeek Moment?
Moonshot AI’s Kimi K2, a 1‑trillion‑parameter mixture‑of‑experts model with only 32 billion active parameters, claims state‑of‑the‑art performance, surpassing GPT‑5, Claude and Grok‑4 on a range of benchmarks including the demanding Humanity‑Last‑Exam test. The model features a 256,000‑token context window, tool‑use interleaving, and...
Tiny 7M Model Beats Massive Counterparts in Reasoning
Tiny 7M Model that Surpassed DeepSeek-R1, Gemini 2.5 Pro, and o3-mini at Reasoning on ARC-AGI 1 & 2 🧠 Last week, I attended @jm_alexia's talk at Mila and I don’t think anyone expected a 7-million-parameter model to outperform models 10,000× larger. Alexia...
Plain-English Guide to How LLMs Really Work
Excited to share our new guest post with @systemdesignone on the System Design Newsletter: “Everything You Need to Know to Understand How LLMs Like ChatGPT Actually Work — 33 Concepts, No Math.” If you’ve ever felt LLMs are “magic,” this is your...
Overthinking Hurts LLMs: Intuition Beats Chain‑of‑thought
What if thinking harder actually makes LLMs worse? 🧠 A new paper by Tom Griffiths shows that both humans and AI models can perform worse when forced to reason step-by-step. In tasks like face recognition or grammar learning, “thinking out loud”...
When Less Thinking Makes AI Smarter 🤯
A recent paper by Tom Griffith finds that prompting large language models to engage in explicit reasoning—often called "thinking" or chain‑of‑thought—can actually lower performance on a range of tasks compared to direct answers. The phenomenon mirrors Kahneman’s System 1 versus System 2...
RAG Boosts Memory, Fine‑tuning Sharpens Judgment
Most people still confuse RAG and fine-tuning — and it’s costing them weeks of wasted effort. Here’s the simple truth: RAG gives your model better memory. Fine-tuning gives it better judgment. RAG connects your model to an external brain so it can retrieve...