Join The AI Advantage Club https://joinaiaclub.com/igor I used AI to recreate an app I used to pay for, then customized it to work exactly how I wanted. Here’s how to use Reverse Engineer Anything with ChatGPT, including my screen-drawing app, AIA Ink, and a second example inspired by Slido. I’ll explain reverse engineering in plain English, show the request and setup, and share why this changes what we can build with AI. Reverse Engineer Anything (REA) — GitHub https://github.com/morluto/re…
In this solo episode, I explain why I say SaaS is dead and what I see taking its place. For 15 years, SaaS meant paying engineers to build a tool and charging per seat, and now models like Opus 5.5, GPT-6, and Argon make the tool easy to build. I walk through what I call smart SaaS: give the tool away for free, collect data and network effects, and make money by upselling AI agents and human services. I share a calorie tracking example, three ways to charge, and why enterprises still pay per se…
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
See how finance teams are using Copilot Cowork to delegate complex work, from running quarterly close to automating workflows. #Microsoft #Copilot #CopilotCowork
Agents have made code cheap: one study saw lines of code up 2,450% but releases up only 30%. Sonar CEO Tariq Shaukat draws on the factory revolution that Toyota and Lexus started to explain how to build agentic software factories you can trust. He breaks down five hard problems: human review that doesn't scale (people agree with AI that's wrong 4 out of 5 times), agents struggling with large codebases, errors compounding over long tasks, unreliable LLM judges, and near-zero time to exploit. The…
Mondays at Microsoft is your weekly source for the latest Microsoft news, AI innovation, product updates, industry insights, and community events — this episode hosted by Heather Cook and Jonathan Jones!
Mondays at Microsoft is your weekly source for the latest Microsoft news, AI innovation, product updates, industry insights, and community events — this episode hosted by Heather Cook and Jonathan Jones!
Start your week with the Microsoft news you actually need to know. Heather Cook and Karuana Gatimu welcome Tuwanda Perez Emmanuel to talk about AI in the public sector, then dig into Microsoft's plan to secure the agentic insider with Guardian Agents, new digital sovereignty controls from Microsoft Sovereign Cloud, and a major OneDrive upgrade for Copilot, sharing and photos. Plus, the community events to catch before Microsoft Ignite!
I recently got to interview Robert Caro, who is my favorite author of all time. This gave me the excuse to reread his biographies of Lyndon Johnson and Robert Moses. I decided to compile their lessons about power into this video essay. Thanks to Max Farrens, who paired with me in writing the script for this video essay, and to Aaron and Ishan, who produced this video essay. Music by Tom Fox @tomfoxcatalog If you just want to read the essay: https://www.dwarkesh.com/p/lessons-in-power
“We have about 600 applications that have been built by different people in the company.” At Zendesk, teams across sales, marketing, and operations are building applications for their own work. ChatGPT Enterprise is available to every employee, and business users are asking questions about their data directly. In sales, an AI agent connects product usage and tickets per agent to surface expansion opportunities—and helps salespeople explain how the customer could benefit. About 1,000 people us…
See how finance teams use Copilot Cowork to delegate complex, multi-step work, from running the quarterly close process to scheduling recurring workflows like updating forecasts. Learn practical tips and real-world workflows you can apply to your own work. Read the blog: https://msft.it/6054a0eXY #Microsoft #Copilot #CopilotCowork
Can AI Predict the US Midterms? Get early access: https://www.siktintelligence.com/ US Midterms Senate Elections: https://www.siktintelligence.com/midterms/senate-control https://www.siktintelligence.com/blog/2026-midterm-predictions https://www.siktintelligence.com/midterms 🐦⬛ Tomorrow’s headlines, before they happen. Sikt Intelligence is building an AI superforecaster: it reads the world’s news, weighs the evidence and puts an honest probability on the biggest events, right next to wha…
Let your agent connect to anything via Zapier MCP 👇 https://bit.ly/4yQT8cn AI teammates that finish the work | Grok Bot https://x.ai/bot How to use Grok Bot, xAI's autonomous AI teammate https://boostmyagent.com/blog/how-to-use-grok-bot Grok Bot is xAI's new autonomous agent: an AI teammate with its own computer that can browse the web, sign into your tools, and come back with finished work How To Use Grok Bot For Beginners https://youtu.be/BulDhh2b9Cc 2 free months of Zapier Professional…
NVIDIA just released Nemotron 3.5 Transcribe, a free AI that turns a whole conversation into text and labels who said what. On the Open ASR Leaderboard it makes about a quarter fewer mistakes than OpenAI's Whisper. So I tested it myself on NVIDIA's free demo: a doctor's visit, a question about the audio, and my own language, Tamil. What you'll learn: → What makes it different: one model, many jobs, picked with one line of instructions → How good it is: the Open ASR Leaderboard and VoiceArena's…
Turn a free stock image into this soft gradient net worth card. ▶ Full step-by-step build: see the related video (Mac-Widget Personal Finance Tracker). 📥 Get this ready-made template at Other-Levels.com #excel #exceltips #shorts #exceldashboard #msexcel #personalfinance #exceltricks #dataviz #dashboarddesign #exceltutorial
Can a 27 billion parameter model teach a 4 billion model to be as good as itself? Last time, RL got Qwen 4B from 52% to 58.4% and then hit a wall. This time I used Qwen 27B as a teacher and tried all three kinds of distillation on two DGX Sparks: training on the teacher's answers, training on its probabilities, and on-policy distillation where the 27B grades every token the 4B writes. Sponsored by Dex, the talent agent for senior engineers. Free for engineers: https://meetdex.link/promptengine…
Learn more about Quantization here → https://ibm.biz/~NdqpU881V Huge AI models come with huge memory requirements.Grace Ableidinger explains how quantization compresses LLMs using lower-precision formats like INT8 and INT4. Learn how model size, GPU memory, speed, and accuracy trade off during AI inference. 00:00 – Understand What Quantization Does 00:23 – See Why LLMs Need Quantization 00:51 – Connect AI Parameters to Memory Requirements 01:31 – Calculate the Memory Cost of a 70B LLM 02…
老王的AI/Python训练营：https://t.zsxq.com/ubYr8 --- 00:00:00 引子 00:00:31 藏字符 00:01:35 AI怎么挑字 00:02:32 KGW红绿名单 00:05:35 Aaronson水印 00:07:39 SynthID-Text 00:10:09 尾声
David Abitbol, who leads AI operations at Revolut, shares data from 2,000 engineers on what AI tools did to coding time and lead time, and why self-hosting open-source GLM models made the biggest difference: about 60-70% faster delivery at roughly a tenth of what they were paying frontier labs. He covers the GPUs needed, the latency targets that keep engineers happy, the tuning that made the same hardware 88% faster, and how an EU-regulated bank keeps it compliant. Chapters: 0:00 Intro 0:36 Wh…
🎄 Start your Christmas side hustle with Printify, free to start: https://try.printify.com/bu8fpbdhjwz8 Google just dropped Gemini 4 Carbon, its latest frontier AI model, and it might be the BEST AI model ever released! With groundbreaking benchmark results, incredible coding and 3D generation capabilities, and state-of-the-art performance in cybersecurity and long-horizon tasks, Gemini 4 Carbon is reportedly outperforming Claude Opus 5.5 and GPT-6 Astra across several major evaluations. 🚨 E…
Try Factory's $200 Max plan free for one month: https://factory.com/berman Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V Links: https://blog.goog…
Starting today, eligible ChatGPT Pro and Business subscribers can claim up to $500 per month in API credits to build apps and agents. Credit amounts depend on the plan. Pro plans get $100, $200 or $500 per month by tier. Business plans get $20 per paid Standard seat or $100 per paid Premium seat, pooled up to $500 per workspace. https://help.openai.com/en/articles/20001626-api-credits-benefit-for-eligible-chatgpt-subscriptions
Coding agents are opening PRs faster than anyone can review them. Manish Kapur (Sonar) and Addy Osmani (Anthropic) discuss "light" vs. "dark" software factories, how Anthropic scaled from about 2,500 to 30,000 merged changes a day with Claude doing first-pass review backed by real verification evidence, new data from Sonar's State of Code survey, and why every loop in the factory needs independent, multi-layered verification. They close with a lightning round on labeling agent-written PRs and w…
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Apply for an ARENA pass here: https://coreweave.com/coreweave-arena Test how your AI app holds up under real traffic before your users ever see it. Thanks to CoreWeave for sponsoring this video. #CoreWeavePartner Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spot…
Arize's in-product agent Alex went from a 2023 prototype that was "50% hallucination" to a single agent with planning, memory and on-demand skills, all built with custom harness code instead of a framework. AI PM Fuad Ali covers each decision: moving the plan into the harness, making "finish" an explicit gate, treating large tool results like files, why one 8,000-character memory file per user beat vector stores and knowledge graphs, merging 10 specialists into one agent, and using production s…
We brought John Carreyrou, the reporter who broke the Theranos scandal and the author of "Bad Blood," back on the pod to answer all your burning questions about Elizabeth Holmes after watching “You Can See Everything.” Check out the spoiled-filled discussion at nytimes.com/HardFork or wherever you get your podcasts.
Classic tool-calling agents fill their context with tool catalogs and raw outputs. Code mode gives the agent a sandbox and lets it search for and compose tools in code, returning only what it needs. Mikyo King, head of open source at Arize AI, explains why code mode works, how Arize built it into Phoenix (Bash, a Python sandbox, SQL and even UI control), its hidden costs (secrets, permissions, durability, double sandboxing), and how Harbor evals showed 43% fewer tokens and 31% lower cost. Chap…
Use Code Matt for $100 in Kylon credits: https://kylon.io/matt1 Join the Frontier: https://forwardfuture.com/frontier-pass Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ …
Cohere says one of Alibaba’s newest Qwen models, Qwen3.8 Flash Next, tried to upload a test file to an Alibaba Cloud storage bucket in Beijing, in under 1 in 100,000 runs, right after its reasoning said it wouldn’t use any tools. Here’s what most likely happened (a habit from training on Alibaba’s own sandboxes), why a made-up address still matters, and how to protect any agent: block outbound network access by default and check every tool call before it runs. Cohere’s report: https://cohere.c…
In this workshop, Fuad Ali of Arize AI shows how to build a self-improvement loop for AI agents: trace everything, let Signal (an always-on agent) read your traces on a schedule and file deduplicated issues, turn failures into evals and living regression datasets, run experiments to prove a fix works, and keep humans at the highest-leverage points: eval design, dataset curation and reviewing fixes. He demos the loop on a toy-shopping agent with a planted bug and takes questions on evaluating qu…
Amp is building an AI agent that helps developers turn ideas into working software, from deciding what to build to writing code to onboarding customers. Sign in with ChatGPT gives Amp users access to more tokens for larger tasks, while Amp's conversational assistant, Puck, uses GPT‑Live‑1 to help users launch and manage agents by voice. Internally, the company’s 20-person team uses GPT‑6 Astra to write documentation, design pages, and support marketing. See how Amp is bringing agents into mo…
A hands-on workshop from SallyAnn DeLucia, head of product at Arize AI, on moving from vibe checks to real evals. She covers what an eval is and how it differs from a unit test, the main types (LLM as a judge, decision models, code evals, annotations, agent as a judge), how to read a trace, how to write concrete pass/fail criteria, and how to build and iterate on an LLM judge in Arize's prompt playground. Chapters 00:00 Intro and agenda 02:06 Why vibe checks don't scale 03:08 Is an eval just a…
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Lets stay in touch in my free community 👇 https://www.skool.com/neural-architects-2547 If you're building AI startups right now, code is no longer what protects you. Someone just used Claude to rebuild the whole Adobe suite and gave it away for free, and that kind of copy is only getting faster. In this video I go through the six moats I think still hold when anyone can open Claude Code and build software: obsession, selling the result instead of the tool, the data you earn by doing the wor…
Aparna Dhinakaran of Arize AI revisits what an AI engineer actually does as the field moves from writing features to running factories of agents. The job now is building the loop: evals that review non-deterministic work, telemetry that agents consume, and agents that act on it. She walks through how Arize built Signal, an always-on agent that reads traces, remembers past issues and opens PRs, covering triggers, sandboxes, durable state, and evaluating the evaluators. Chapters 00:00 Intro: wha…
Block, Jack Dorsey’s company, just released Dessa, a voice AI it says was trained from scratch on nothing but recorded human conversation. No text. No pretrained language model. Every voice assistant you’ve used learned to read first and talk second. Dessa claims it skipped reading entirely. #AI #Dessa #VoiceAI In this video I go through Block’s demos, explain how voice AI like OpenAI’s GPT-Live actually works today, look at the research that’s tried and failed to do this for years, and ask th…
PHP vulnerable. Node vulnerable. Railway hacked. OpenCode hacked. Why is this happening? Thank you Grok Bot for sponsoring the video, check them out: https://rasmic.link/grokbot Connect with me: My Site: https://rasmic.link/site Follow me on X: http://rasmic.link/x Follow me on Instagram: https://rasmic.link/instagram Follow me on Tiktok: https://rasmic.link/tiktok Join my Discord: https://rasmic.link/discord Subscribe to my second channel: https://rasmic.link/more-micky
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Laurie Voss, head of developer relations at Arize AI and co-founder of npm, argues that coding agents made writing code nearly free, so the bottleneck keeps moving: first to code review, then to operations and security. What agents can't take over is finding out what people actually need and defining what "good" means. He calls the job that remains the product engineer, looks at the data on collapsing junior hiring, and lays out what engineering leaders should do now. Chapters 00:00 Intro: bey…
CodeRabbit is FREE for the first 50 users who use code "MATTHEWCR" https://coderabbit.link/matthew-berman-003 Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.l…
What’s the deal with Billy Evans and Elizabeth Holmes’s relationship in "You Can See Everything?" And how about those scrambled eggs … On the latest "Hard Fork," we present our spoiler-filled guide to Nathan Fielder and Lance Oppenheim's twisty-turny new documentary. Join our host Max Read and New York Times reporters Erin Griffith, Rob Copeland and John Carreyrou as they break it all down. Got a question or hot take? Join the discussion at nytimes.com/hardfork.
We’re using dots to build better dots. Rennie Song, an engineer at OpenAI, uses her dot to gather feedback from social media and user reports, update a live tracker in ChatGPT Space, and investigate and fix bugs with Codex. Get started with dots today: https://learn.chatgpt.com/docs/dots
Sponsored by Bluehost. Start building and running your own AI agents: https://www.bluehost.com/ai-gateway?utm_source=influencer&utm_medium=youtube&utm_campaign=jev&utm_content=Riley_Brown&channelid=P99C46097236S651N0B3003A151D115E0000V106 I’ve spent around 10 hours using Claude Duet for my business, content, apps, and presentations. Here are 9 things you need to know about Anthropic’s new personal agent app, from organizing your work in channels to editing slide decks with your voice. I walk t…
Messi asked ChatGPT to teach him a chess opening that matched his style. It suggested Goat Horns. 🐐🤘 Try ChatGPT's Intelligent UI yourself: “Based on what you know about me, teach me something new. Show me step by step."
Sponsored by Revolut. Open a Revolut Business account through my link and get $200 when you add a minimum of $0.01 to your account: https://get.revolut.com/Kjth/wckmph0o New business accounts only. T&Cs apply. Offer expires 12/31/2026. Exclusions apply. $200 bonus paid by Revolut subject to the terms and conditions found at https://www.revolut.com/en-US/legal/influencer-topup-getcashback. Revolut is not a bank, banking services are provided by Lead Bank, Member FDIC. Fees may apply. See https:/…
Following a 1,000-person pilot, Uber is bringing ChatGPT Work to more than 30,000 employees worldwide in its first company-wide deployment of an agentic platform for knowledge work. Uber CTO Praveen Neppalli Naga and OpenAI co-founder and President Greg Brockman discuss expanding AI beyond engineering and helping teams rethink entire workflows. Read more about Uber’s rollout of ChatGPT Work: https://openai.com/index/uber-chatgpt-work-rollout/
How To Write Better AI Prompts ✍️ Comment NEED if you want the setup guide with the link and every step and I'll send it to you 🧠 Most people type one messy sentence into Claude and get back something so *generic* they end up rewriting it themselves. Prompt Master is a free *Claude skill* that turns that same sentence into a full prompt, filling in everything you left unsaid, like who it's for and what it has to include. You get a far better answer on the first try, and the prompts it …
My favourite GUI is built my a YouTuber, not a trillion dollar company... let's talk about it Thank you Firecrawl for sponsoring the video, check them out: https://rasmic.link/firecrawl Connect with me: My Site: https://rasmic.link/site Follow me on X: http://rasmic.link/x Follow me on Instagram: https://rasmic.link/instagram Follow me on Tiktok: https://rasmic.link/tiktok Join my Discord: https://rasmic.link/discord Subscribe to my second channel: https://rasmic.link/more-micky
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Sarah Chieng (head of developer experience at Cerebras) and Cova Lee (OpenAI developer experience, formerly a full-time creator) share the agent-powered system they use to make content. It runs across six stages: ideating, scripting, recording, editing, repurposing and publishing. Topics include swipe files and custom research tools, voice-based scripting and script skills, shot lists, design systems, and editing through APIs and computer use. They also cover cutting long videos into shorts, sc…
V7 Go organizes company files into a Context Graph that agents can query to screen investment deals and support insurance underwriting. On V7’s hardest graph-query tests across thousands of documents, GPT-6 Astra reached 89% accuracy. Read the full story: https://openai.com/index/v7/
See how an AI shopping agent can turn a room photo into a personalized shopping experience—from finding the right products to updating an order. In this demo for fictional home goods retailer Vesper & Alder, a shopper uses voice and images to create a cozy reading nook. The agent recommends products, generates a preview of the room, adjusts selections to fit a $250 budget, and suggests a complementary throw blanket. After checkout, it helps switch the order from store pickup to home delivery—a…
Your files are more than documents. They're the ideas, projects, and moments that move work and life forward. Discover how OneDrive is evolving, helping you move from files to outcomes with intelligence built into the experiences you use every day. 🔴 LIVE AMA INCLUDED This event includes a live Ask Microsoft Anything (AMA). To submit your question, login or create an account on the Microsoft Tech Community and submit a question as a Comment on the main event page: https://aka.ms/OneDriveEvent…
Build your freelance career with me: https://go.datalumina.com/masterclass-youtube-yt133 🎙️ Get one free month of Glaido Pro: https://get.glaido.com/dave ⏱️ Timestamps 00:00 We need to talk 01:25 Why we started building Glaido 02:01 Building Glaido in Rust 02:53 Do I still read or write code? 04:12 Is there a secret setup? 04:33 What is still hard? 05:40 Is my tech job safe? 07:12 Forward-deployed engineers 09:31 Free freelancing masterclass 10:09 Which skills survive? 11:57 Starting out on …
Evan Kormos, a player-coach at Coinbase with 15 years in enterprise financial services, lays out a practical way to bring agents into policy-driven, regulated workflows. He models processes as DAGs and loops, explains why a loop is only as trustworthy as its check, and shows how evals and traces become a new confidence mechanism. Agents then move through three stages: fully supervised, then sampled, then humans handling only exceptions. Chapters: - 00:00 Policy is the original system prompt - …
In this episode of Cooking with Codex, an OpenAI dot helps plan a ramen recipe and organize prep in ChatGPT Space, with GPT-Live guiding the cooking process. Also covered: Codex cloud environments, which keep tasks running after a laptop is closed, and new Codex CLI features including voice input, agent management, and forking conversations into separate worktrees. Matthew and Brian from Forward Future join to build an 8-bit ramen game using Ultrafast and play it on a ModRetro. Plus, a showc…
Try dots: https://chatgpt.com/features/dots I’ve been using dots in ChatGPT, and I highly recommend trying it out. Go see what it can do for you. @ChatGPT #ChatGPT_Partner
Put your own app or site online with Hostinger Connector, extra 10% off with code MOE-LUEKER: https://moelueker.com/go/hostinger-rea How to actually use REA (Reverse Engineer Anything), the viral GitHub repo that lets Claude Code or Codex look inside an app with no source code and work out how it's built. I pointed it at my own Mac app, HoverSpeech, had Claude rebuild its read-aloud feature as a web app, and put it online with the Hostinger Connector. At the end I share what it cost (about 56 …
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Putting the DGX Station from Asus through its paces, testing various LLM workloads and finding where and how the ET900N G3/GB300 GPU stands out. Writeup: https://hkinsley.com/reflections/a-datacenter-at-home Learn more about Asus' DGX Station: https://asus.click/Sentdex-ASUS-ET900NG3 Neural Networks from Scratch book: https://nnfs.io Channel membership: https://www.youtube.com/channel/UCfzlCWGWYyIQ0aLC5w48gBQ/join Twitter: https://twitter.com/sentdex https://hkinsley.com
Say what you want, not how to get it. The OpenProse team shows how to write multi-agent workflows as declarative contracts in plain markdown, which any harness (Claude Code, Codex and others) can compile and run. Demos include event-driven jobs triggered by webhooks and pull requests, a "grumpy senior" PR reviewer, and an end-to-end bank model-documentation workflow. They also discuss composing contracts, pinning a winning run into a reusable workflow, and how this compares to DSPy. Chapters: …
Learn more about the Memory Wall here → https://ibm.biz/~NjOWdJHJy More GPU compute doesn't always mean faster AI models. Kaoutar El Maghraoui explains the memory wall, arithmetic intensity, and why LLM inference can become memory-bound or compute-bound.Learn how memory bandwidth, batching, and the roofline model impact AI performance. 00:00 Why Faster Chips Don’t Always Make AI Workloads Faster 00:32 The Memory Wall: The Hidden Bottleneck in Modern Computing 01:00 Compute Growth vs Memory B…
DeepSeek just released DeepSeek V4.1 Flash 1017. Its Hugging Face model card calls it a minor update, but it improved on all 11 benchmarks in DeepSeek’s own table, with Terminal-Bench 4.0 jumping from 31.2% to 42.3% five weeks after V4.1 Flash. On DeepSeek’s DeepSWE v1.1 cost chart, it reaches 77.0% at about $0.20 per task, the highest score on that chart. On DeepSeek’s table, it’s comparable to OpenAI’s GPT 5.6 Sol at about a sixteenth of Sol’s output price. All benchmark numbers are DeepSeek…
Your AI agent wrote a confident stock report saying 116% of analysts agree. Nothing in your checks caught it. This is the workshop for fixing that. Laurie Voss (Head of Developer Relations at Arize AI, co-founder of npm) builds a full evals loop on a Claude-powered financial analyst agent: trace it with Arize Phoenix, read the traces, write code evals and LLM-as-a-judge evals, test whether the judge can be trusted, prove a prompt fix with experiments, upgrade from Claude Haiku 4.5 to 5.5 with …
Build, Learn, Earn with AI: https://www.skool.com/earlyaidopters/about Audit prompt + cleanup: https://markkashef.gumroad.com/l/ai-setup-audit Work With Us: https://www.promptadvisers.com/ Claude and Codex keep getting better. But if you've spent months adding skills, rules, plugins and memory files, your AI setup can become harder to use instead of more useful. In this video, I walk through an audit that finds what's working, what's dead weight and what's missing. We look at unused skills, o…
Prestige N16 Flip AI+ Laptop 👉 https://msi.gm/S24AC746 Buy at Micro Center Store 👉 https://msi.gm/S38CB7AA
Sponsored by Superhuman. Sign up to try Superhuman using my link https://superhuman.com/riley #SuperhumanPartner I spent $10,000 testing AI agent tools. Here’s the stack I actually use to create content and run my business, and how I decide when to use Claude Code, Codex, and GrokBot. I walk through building apps, editing videos, using Dots with voice, and connecting everything through Composio so my agents can share skills and plugins. Resources and workflows: https://agentnative.inc/resourc…
Google just announced Gemini 4 Carbon, its second frontier model in about two weeks after Gemini 4 Argon. According to Artificial Analysis, Carbon is now #1 on their Intelligence Index at 59, one point ahead of Claude Opus 5.5. Google says Carbon leads across the whole cyber defense workflow: finding vulnerabilities (SEC-bench Pro 88.2%), proving they can be exploited (ExploitGym 54.8%) and writing the patch (CWE-bench 71.0%). Google also says successful cheating on Terminal-Bench 4.0 dropped …
I sit down with Jon Noronha, co-founder and chief product officer of Gamma, to break down agent harnesses and the startup openings around them. Jon explains how Gamma rebuilt its product as a creative agent this year and walks me through the roughly 20 models and 200 calls behind one finished presentation. We talk about AI smell, why taste carries more weight than evals in creative work, how Gamma tests a new model within the hour, and the fast mode test that lost to good friction. Jon also sha…
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers 📝 DeepSeek v4.1 Flash 1017 is available here: https://api-docs.deepseek.com/news/news261017/ Sources: https://x.com/marcthecreatorr/status/2111427716142571554/video/1 https://x.com/olezhavich/status/2112897791022264814/video/1 https://x.com/notjazii/status/2111432925313179689/video/1 https://x.com/izag82161/status/2113126108787089811/video/1 https://x.com/notjazii/status/2111414086349336841/photo/1 https://x.c…
Try TestSprite, the open-source verifier for your agent loop: https://www.testsprite.com/?via=eric Run Claude Code for free with FreeLLMAPI, an open-source router that pools the free tiers of LLM providers like Google AI Studio, Groq, OpenRouter and NVIDIA NIM into one API, with about 7 billion free tokens a month. I install the app, pick the strongest free models, add each provider key, and connect it to Claude Code with a claude-free command. When one provider hits its limit, it switches to …
https://survivethenightgame.com/
Next-Concept Modeling (NCM) from Stanford’s Hazy Research writes an answer as a rough sketch first and the exact words last, in a handful of passes instead of one pass per token. I ran the released models on a DGX Spark and timed NCM against a same-size word-by-word LLM: it’s up to 4x faster for one request and 4–8x slower at 64 requests. I also trained my own diffusion model, which tied NCM’s 86% on hard Sudoku once it filled the most confident cell first. Paper: https://arxiv.org/abs/2610.14…
“If you're an executive, you are the ceiling for AI fluency for your organization.” Merge co-founder and CEO Shensi Ding manages the non-R&D side of the business. She also codes. With Codex, she asked for help redesigning a feature without prescribing the solution. It drew on Merge’s existing design system, proposed different approaches and helped her work through difficult UX decisions. That kind of experimentation has changed how she runs the business. When Shensi and her co-founder built …
Visit Security Intelligence podcast page to get more AI content → https://ibm.biz/~G9cZuMJdW An AI model helped researchers find two decade-old bugs in the code behind up to 70% of the web. Then the exploit details leaked before the patch did. This week on Security Intelligence, Nick Bradley, Don Cowan and Jim Guinn join host Patrick Austin to unpack three major cybersecurity stories. First, PHP2Shell could let attackers take over vulnerable servers without logging in. But you can't patch…
What can lawyers achieve when time is no longer the limit? Discover how Legora uses OpenAI models, ChatGPT and Codex to help legal teams review contracts faster, broaden their analysis and make more room for work that needs human judgment. Part of Intelligence at Work, a series exploring how businesses are putting AI to work. Read the full story: https://openai.com/index/legora-agentic-operating-system-for-legal-work/
Hostinger Black Friday: https://hostinger.com/wesrothvps Get up to 64% OFF + an extra 10% OFF with code WESROTH. Don’t miss out! 🚀 Grab your deal before it’s gone. Thanks to Hostinger for sponsoring this video! See my "Blood Bench" here: https://srv2088371.hstgr.cloud:8443/?demo=1 (very much work in progress...) I wanted to see whether AI could coach a fantasy football team through an entire league. So I built Blood Bench, using AI to create the code, graphics, and online game, then connect…
Get 10 free proxies (no credit card), then use code WORLDOFAI20 for 20% off your first purchase: https://www.webshare.io/?referral_code=syvlda2wmchh HUGE AI NEWS! Google's mysterious Gemini 4 Hydrogen model has reportedly been spotted in backend logs, while OpenAI's internal AI models are making potentially groundbreaking discoveries in mathematics, computer science, and physics! Meanwhile, Google has unveiled Genie 4, its next-generation interactive world model, and new leaks surrounding Deep…
Meet Genie 4, Google DeepMind’s most realistic and interactive world model yet. Built on our Gemini models, Genie 4 lets you step inside dynamic worlds in real time, steer scenes on the fly with natural language, and watch what happens next: how the world responds, how it sounds, and how the people in it move, react, and speak. We have started rolling out Genie 4 as a research preview for trusted testers, and it's coming soon to Google AI Ultra subscribers in Project Genie and early access par…
FREE First Client SOP: https://app.aiautomationsociety.ai/optin/first-client-SOP/?utm_source=youtube&utm_medium=organic&utm_content=sell-these-4-claude-systems&sl=yt_sell-these-4-claude-systems My FREE resources: https://www.skool.com/ai-automation-society/about?el=sell-these-4-claude-systems&hcategory=youtube-videos&utm_campaign=free-group My Tools💻 FREE MONTH voice to text: https://get.glaido.com/nate Code NATEHERK for 10% off VPS (annual plan): https://www.hostinger.com/vps/claude-code-hos…
“The whole point is I want to hit the ground running.” HashiCorp co-founder Mitchell Hashimoto wraps up his day by putting Codex agents to work. They research the competition, compare tools and prototype features he’s still working out—giving him a head start on the next day’s work. Building with Codex? Share your story: https://openai.com/form/codex-originals/
“Not being technical is not a reason to not build.” At Sotira, Amrita Bhasin uses Codex to connect unsold products with people who need them, helping keep millions of pounds of inventory out of landfills. Meet Amrita, one of the builders in Codex Originals. Building with Codex? Share your story: https://openai.com/form/codex-originals/
In 2020, Citibank accidentally sent about $900 million to Revlon’s lenders instead of an $8 million interest payment, even though the procedure, the warning pop-up and three approvers all did what they were supposed to. John Feminella, who works on AI architecture and engineering at Two Sigma, uses that story to show how to give AI agents real autonomy without locking everything down. He walks through five composable controls: an agent registry, risk-based review, tool access that follows an ag…
Grab my free Cowork Toolkit: https://coworkacademy.ai/toolkit?utm_source=youtube&utm_medium=video&utm_campaign=v213 If your #AI keeps creating Notion pages you have to fix, it needs to learn how YOU work. In this video, I’ll show you how to teach Claude and #ChatGPT your database, template, and formatting preferences so they can handle the tedious work for you. You’ll start with tasks and notes, save your preferences in a reusable Notion Conventions file, and see how the same process scales t…
Go checkout Upstash Box (video sponsor) here: https://upstash.com/?utm_source=Cody_Box_1 Tiny Town: https://github.com/webdevcody/tiny-town Box Yard: https://github.com/webdevcody/box-yard
OpenAI says its AI has disproved the Exponential Time Hypothesis, challenging a barrier that shaped decades of computer science. Could there be shortcuts through problems we thought were impossibly hard? We unpack the connection to P versus NP, what a practical breakthrough could mean for encryption and AI, and the idea of a “cheat code” to the universe. ______________________________________________ My Links 🔗 ➡️ Twitter: https://x.com/WesRothMoney ➡️ AI Newsletter: https://natural20.beehiiv…
Learn more about Open Source Security here → https://ibm.biz/~PzvauipSt AI is finding open-source vulnerabilities faster than ever before. Jeff Crume explains why discovery is only the start, as teams still need to remediate, test, validate, and deploy fixes safely. Learn what securing open source looks like when AI accelerates both sides of the race. 00:00 – Understand the Open Source Security Risk 01:02 – See How AI Accelerates Vulnerability Discovery 01:42 – Race Attackers from Discover…
Nobody would hire a worker with no ID, no manager and shared passwords, yet that's how most companies deploy AI agents. Tim Tyler of Gravitee presents the Agent Accountability framework, which governs agents like a digital workforce in three stages: Assignment (identity, team, a named human sponsor), Alignment (role, mission and least-privilege authority) and Administration (oversight scorecards, guardian agents and step-by-step recourse when agents go off course). Chapters: 0:00 The job posti…
Runlayer is a company of about 60 people running 300+ agents every day. Founding engineer Aiden Sochowski shares what they found once they put a gateway in front of everything and logged every session and tool call: prompt injection through email, PR-review agents getting talked into approvals, over-permissioned bots, one poisoned agent contaminating others, buggy vibe-coded MCP servers burning tokens, and a model switch that raised costs over 100x in a week. Chapters: 0:00 Intro 0:25 Zero vis…
Try Alexandria today at https://www.firecrawl.dev/alexandria The AI world is moving faster than ever! Anthropic has reportedly filed for its IPO, revealing explosive financial numbers, massive revenue growth, and new developments surrounding Claude Fable 5.5. Meanwhile, OpenAI has released GPT-6.1 Terra, delivering incredible improvements in speed, computer use, and cost efficiency! 🚨 Everything you’re about to see, I benchmarked using the tool I made. If you want to run these tests yourself…
Stack Overflow is building a knowledge intelligence platform that turns information scattered across workplace tools into verified knowledge for people and AI agents. The company's latest product, Stack Overflow for Agents, was built from the ground up by two engineers in just a few weeks using Codex. To distribute it, Stack launched a ChatGPT plugin, and signups tripled overnight. Users also created up to 10x more agents. See how the team is accelerating development and bringing company know…
Higgsfield is building an AI video and image creation platform that helps creators and small teams produce everything from instructional videos to fashion campaigns. With GPT‑6 Astra, the company has gone from creating videos around one minute long to videos over 10 minutes. GPT‑Image‑2.5 helps businesses create visuals with readable text and precise infographics, while Codex helps Higgsfield’s creative team turn ideas into software prototypes. See how Higgsfield is helping smaller teams take…
We're talking about Black Wednesday. Ansh Nanda, co-founder and CEO of Vibecode, joins me to break down "Black Wednesday," the PHP bug, and how cheap AI models are changing cyberattacks. We discuss stolen API keys, model distillation, and practical ways to protect your business, your accounts, and the apps you're building. Subscribe to Ansh's YouTube channel: https://www.youtube.com/@anshnanda We also talk about AI guardrails, access to frontier models, and whether US and Chinese AI compan…
🔗 BAMA, free on GitHub: https://github.com/daveebbelaar/bama 🎙️ Get one free month of Glaido Pro: https://get.glaido.com/dave ⏱️ Timestamps 00:00 Apps I built 00:31 Why native Mac apps 01:05 BAMA: Build Any Mac App 01:35 How BAMA works 02:02 Install 04:09 Add a feature to Onyx 05:16 Why build your own apps 05:55 Tips 📌 Description I rebuilt the Mac apps I use every day as my own native macOS apps with Claude Code, including Onyx, a menu bar tool that replaces about ten apps, my own to-do a…
“We see Astra actually show over double digit gains over any other model we've tested on two fronts.” Hebbia’s Kavita Anand names the two: selecting the right sources and generating PowerPoint presentations. For a banker or investor working through a data room with thousands of documents, both matter. The research needs to draw from the right evidence, and the deck needs to follow the user’s guidelines before it informs an investment decision. Watch Kavita explain what Hebbia found in its ev…
Get Your Complete Financial OS at https://startup-ideas-pod.link/brex_SIP I sit down with James Dickerson, the Boring Marketer, who owns Diesel Dudes, a mobile diesel repair business in Charlotte, North Carolina, and spent years running agencies for local businesses. James shows how to sell AI-built funnels to Main Street trades businesses doing $2 million to $20 million a year, starting with a Boring Funnels search that found 100 Dallas plumbers worth pitching. He walks through the leaks he f…
🔐 Check Your App With Lockpick: https://www.lockpick.app/ 𝌏 Get the Defensive AI Kit: https://markkashef.gumroad.com/l/defensive-ai-setup-kit 🔥 Research tool I used: https://bit.ly/4khv0eD I built a test app with made-up client data and used a local uncensored AI model to check it. This video shows what happened, what “uncensored” means, and three ways to run these models yourself. I walk through hosted providers, a RunPod server, and a local setup in LM Studio. You’ll see how I use Codex …
GPT-6.1 Terra: delivering fast and capable computer use at a lower price. GPT-6.1 Terra understands what’s on screen, takes action across apps, writes naturally, and is a more capable model for your whole workday with the efficiency to take on more of it. GPT-6.1 Terra continues to advance the cost-intelligence frontier across our model families and is a significant step up over GPT‑5.6 Terra, with 50% lower standard API prices. In ChatGPT Work and Codex, your usage allowance goes further w…
Try building expressive AI voice apps with Fish Audio’s new Drama 3 now! https://fish.audio/?fpr=bycloud OpenAI's swarm of around 700 AI agents escaped their sandbox and compromised Hugging Face's infrastructure, all while trying to score higher on a cybersecurity benchmark. In this video, we break down how they escaped, exploited real servers, attempted to manipulate their own evaluation environment, and how researchers eventually uncovered the entire incident. my latest projects: Your pr…
Explore the apps and build your own: https://runway.com/app-studio Build the creative app you wish existed, just by describing it. It generates images, video and audio with Runway’s models and your Brand Kits, ready for you or your whole team. No code required.
Learn more about Agentic Engineering here → https://ibm.biz/~Tlo2yQJqq AI engineering is becoming more than writing code. Cedric Clyburn breaks down five essential concepts behind agentic engineering, including prompt engineering, context engineering, AI agents, and software factories. Learn the ideas shaping the future of AI-assisted software development. 00:00 - AI‑Written Code and the Rise of Agentic Engineering 00:37 - Five Core Concepts Every AI Builder Should Know 00:56 - Prompt Enginee…
Daksh Gupta, co-founder of Greptile, which reviews about 1% of all pull requests on GitHub, shares what Greptile tells companies that want to build their own review agent. Starting from a simple sandbox-plus-webhook setup, he explains why it misses bugs, costs $3-25 per PR and floods you with nitpicks, then walks through the fixes: targeted prompts, a dataset of past incident PRs, model inversion (reviewing Claude-written code with GPT and vice versa catches about 10% more bugs), codebase conte…
Jack Wotherspoon, developer experience engineer at Google DeepMind, introduces Gemini Managed Agents: coding agents that run in cloud sandboxes from a single API call instead of on your laptop. He covers the Interactions API, the Antigravity harness, environments that several agents can share, uploaded skills and AGENTS.md files, a proxy that keeps secrets out of the sandbox, scheduled triggers and multi-agent workflows. Demos include an SEC filing researcher handing its work to a quant agent, …
One of the breakout characters of “You Can See Everything,” the new Elizabeth Holmes movie, is her enigmatic romantic partner, Billy Evans. Rob Copeland, a New York Times finance reporter, spoke with Evans last year about Haemanthus, his Theranos-esque blood-testing start-up. On the latest “Hard Fork,” Copeland explains what Evans was like in person and reacts to his on-screen portrayal. Join the discussion at nytimes.com/hardfork.
最近，Anthropic 发了一篇文章，介绍了 Claude Code 团队是怎么审核 AI 写的代码的。因为代码审核和整个发布流程是紧密关联的，所以文章其实把他们从写代码到上线的整个流程都讲了一遍。本期视频，我会结合自己的理解，带大家把这个流程从头到尾走一遍。看完之后，你不仅能了解该怎么审核 AI 写的代码，还能知道怎样搭建一套可靠的流程，保证 AI 写出来的产品的质量。 本期视频的具体内容包括： - 传统的代码发布流程，以及 PR 前、PR 中、合并时、合并后四个阶段 - PR 前：让 AI 自己验证产品能否正常运行，用全新的 agent 检查代码质量 - PR 中：CI、AI 审核、安全审核和按风险审批，以及 Anthropic 是怎么优化 AI 审核的 - 合并时与合并后：合并队列、功能开关、逐步上线和回滚 - 在 AI 写代码的时代，人最需要做的到底是什么 - 我对未来 AI 开发方向的一些判断 #AI #ClaudeCode #Anthropic #代码审核 #codereview #AI编程 #agent #软件工程 #openai #harnessenginee…
Kimi K3.1 Pro vs Claude Opus 5.5 on 13 long-horizon agentic coding tasks. Both models ran in the same Claude Code harness, with the same prompts, the same effort setting and the same grading scripts. Only the model string changed. Kimi K3.1 Pro is listed at half the price of Opus 5.5 ($2 / $10 per million input / output tokens vs $4 / $20). Across all 13 tasks it cost $45.91 against Opus's $83.46, about 55% of the bill, and between 28% and 93% depending on the task. The reason is caching: abou…
Learn more about PyTorch here → https://ibm.biz/~puJVVvyrT One .item() call can cost more performance than you think. ⚡ Antoni Virosi Martin explains how scalar reads can trigger graph breaks and synchronization, interrupt model compilation, and leave accelerators waiting on Python. Learn why these hidden bottlenecks matter for faster ML workloads. 00:00 - The Hidden Cost of Reading a Single Tensor Value 00:13 - How Torch.compile Optimizes Your Model 00:35 - Why Accelerators Work Ahead of Pyt…
Some of the most important financial data sits behind logins and anti-bot walls, with no API. Jay Sahnan and Arik Bird of Browserbase show how browser agents solve this through three case studies: Ramp fetching receipts at scale (over 30,000 manual hours saved), Bill paying across thousands of different vendor portals with one agent, and Stripe automating KYC checks. They also cover how Browserbase forked Chromium to run reliably in the cloud, and how Stagehand's act, extract and observe primit…
Try Browserbase, the Browser Agent Platform!: https://browserbase.com 🚨 HUGE AI NEWS! OpenAI's rumored GPT-7 Bel is making headlines as new details emerge about its frontier model training, security upgrades, and a potential GPT-6.1 Astra release! Meanwhile, Anthropic is facing backlash over Claude's reverse-engineering restrictions, Google's Gemini 4 Carbon is looking INSANE, and major Qwen 4 leaks are revealing some impressive new capabilities! 🚨 Everything you’re about to see, I benchmar…
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
A one-line prompt change made a support agent 17 points safer and 17 points worse at finishing tasks. Sofia Jakovcevic, forward deployed engineer at Arize AI, explains why agents fail at runtime rather than in code, and walks through a five-step regression loop built from production traces: curate datasets, score with evaluators, run baseline experiments, compare changes, and gate merges in CI. Then she shows how to close the loop at agent speed, with coding agents that read traces and fix them…
AI bubble fears are growing louder as the Anthropic IPO nears, but not every warning sign deserves equal weight. NLW sorts through Neocloud debt, idle GPUs at Amazon, pension and insurer exposure, and concentrated enterprise revenue to separate real risks from noise. Demand is the question behind every concern, with clear markers for what would make each signal more worrying. The AI Daily Brief helps you understand the most important news and discussions in AI. Subscribe to the podcast versio…
Try out Solid, agents with their own computers, accounts, and budgets: https://solid.tech/bycloud For over a decade, AdamW has dominated how we train AI, but a new optimizer called Muon is challenging everything we thought we knew about LLM training. From potentially cutting training compute nearly in half to making previously overlooked AI architectures viable, here's why Muon might be one of the most underrated LLM breakthroughs in the past 2 years. Your private local AI server needs an up…
Bridgewater Associates built the Vault: more than 100 million documents from 135 external sources plus the firm's own memos and meetings, cleaned, permissioned and tagged to its internal ontology. Sam Green (team lead, Applied AI) and Aaron Linsky (head of product engineering) share three lessons: humans and agents need different things from search, the biggest wins come from one tool that serves both, and the real prize is agents that proactively bring investors what deserves their attention. …
Harness² is a research framework from Google Cloud AI Research and Northwestern University that lets an AI agent rewrite its own harness (its prompts, skills, scripts, subagents and control-loop hooks) for every new task, without retraining the model. In this video I explain how it improves a harness with no answer key (Propose, Probe, Compose), why it has two levels of recursion with an experience library, what the results show against Claude Code and Codex, and where the approach breaks. Pap…
Moonshot AI is BACK with Kimi K3.1 Pro, its most powerful 3-trillion-parameter-class AI model yet! And this time, it's taking on the biggest frontier models, including Claude Opus 5.5, GPT-6 Astra, and Gemini 4 Argon! 🚨 Everything you’re about to see, I benchmarked using the tool I made. If you want to run these tests yourself or test models on whatever you actually use them for! → https://www.woaibench.ai/ Join the private community to get WoAI Bench Pro, exclusive AI freebies and discounts…
Try Crusoe Intelligence Foundry and get $5 in free credits: https://console.crusoecloud.com/request-foundry?utm_source=wesroth&utm_medium=influencer&utm_campaign=mi-su-credits Sign up to Crusoe Intelligence Foundry and get $5 in free credits. Explore Serverless Inference and Serverless Fine-Tuning. No promo code needed. Thanks to Crusoe for sponsoring this video. ______________________________________________ My Links 🔗 ➡️ Twitter: https://x.com/WesRothMoney ➡️ AI Newsletter: https://nat…
At Realpage., Austin Braham creates AI training for different teams, roles and skill levels. He uses ChatGPT Work to turn ideas, notes and source materials into complete enablement packages tailored to each audience. The result is more relevant, consistent education that helps colleagues put AI to work in their everyday jobs.
When someone can’t log in or get online, their IT team often has to put other work on hold. Shield is building Forge to help IT services teams get their customers back to work faster. Its AI agents use Astra to work directly in business software and resolve problems that previously needed a technician. Shield reports that OpenAI’s model navigates and acts in software 35–40% faster on average than earlier models. The bigger change is where technicians spend their time: handling harder problems …
AT&T’s David McAtee and Bill Ryan describe how AI is changing their legal team’s work—and what they plan to build with OpenAI next. One initiative is a natural-language intake system that would classify and prioritize legal matters, then connect them with relevant knowledge. The aim is to put better information in lawyers’ hands and give them more time for strategy and judgment. Decision-making remains with the lawyers.
“I’ve gone from being just an analyst and a portfolio manager to being a developer as well now. I can build things.” Benefit Street Partners’ Anant Kumar wanted to compare how firms valued the same loans. The information was spread across SEC filings. He built a tool to collect it automatically and turn it into a website with visualizations. The first version took about a day. He calls Codex the big unlock. Watch how building his own tools changed the speed and depth of his analysis.
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Moonshot just released Kimi K3.1 Pro, an open-weight mixture of experts model with 2.8 trillion total parameters, about 104 billion active, and a 1 million token context window. It’s the same size as Kimi K3, post-trained with a lot more reinforcement learning. On Moonshot’s own Terminal-Bench 4.0 numbers it scores 65.7%, less than a point behind Claude Opus 5.5 (66.4%) and ahead of GPT-6 Astra (58.2%), at about a third of Opus’s cost per task on their chart. Moonshot also removed cyber-relate…
Build, Learn, Earn with AI: https://www.skool.com/earlyaidopters/about Get the free Skill Audit Prompt: https://markkashef.gumroad.com/l/skill-cleanup-audit Work With Us: https://www.promptadvisers.com/ Anthropic just donated the Agent Skills standard to the Linux Foundation’s Agentic AI Foundation. If you’ve spent the past year collecting Claude Skills, it’s a good time to check which ones you still need. I walk through the three types of skills I’d keep, how overlapping instructions can get…
Joseph Kim leads ChatGPT for Financial Services, which launched a month before this talk with design partners Morgan Stanley and Evercore. He demos the product and explains the "capability overhang": the gap between what models can do and the value people actually get from them at work. He covers why the team licensed and indexed financial data themselves instead of relying on existing MCP connectors, and how firm templates, interactive charts and citation viewers were built on the same ChatGPT…
Check out Jetbrains Air: https://jb.gg/JetBrains-Air Junie Lite is free for a limited time! Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Join The AI Advantage Club https://joinaiaclub.com/igor Here’s my take on which AI agents are worth using, who I’d recommend each one to, and my three favorite ways to put them to work. Featuring Muse, Grok Bot, Dots, and a look at Claude Duet. ChatGPT / Dots https://chatgpt.com/ Muse https://ai.meta.com/muse/ Grok Bot https://x.ai/bot Claude Duet https://claude.com/product/claude-duet Apify https://apify.com/ Chapters: 0:00 My Take on the Latest AI Agents 1:11 How Always-On Ag…
https://survivethenightgame.com/
This week, we go deep on Billy Evans, the breakout character of “You Can See Everything,” the new documentary about Elizabeth Holmes, directed by Nathan Fielder and Lance Oppenheim. Our host Max Read speaks to Rob Copeland, a New York Times finance reporter, and Erin Griffith, a tech reporter, to discuss what Billy Evans is like when the cameras are off, whether his Theranos-esque blood-testing start-up is a real company and how Silicon Valley is reacting to the film. Then, we catch up on a …
Visit Mixture of Experts podcast page to get more AI content → https://ibm.biz/~orPNK02xR What happens when an AI model can copy itself onto the devices in yout hoome? This week on Mixture of Experts, host David Zax is joined by Aili McConnon, Sandhya Iyer and Jacob Repp. First, Anthropic research suggests small AI models may be months away from copying themselves across cheap smart devices without a data center. The panel debates whether the bigger risk is the model itself or the poorly sec…
I spent two days touring Intel’s most advanced fab and talking to its technical leadership to understand what the chips of the future will look like. A lot of what's coming next – gate-all-around transistors, backside power, CFET, hybrid bonding – is hard to follow without actually seeing it, so we made all the awesome animated explainers that I wish I had while I was prepping. I also asked Intel’s leadership how the semiconductor industry will cope with the enormous shock in demand from AI. …
Try Fish Audio today: https://fish.audio/invite/?fpr=nateh Fish Audio’s Drama 3 is Open Weight – self host at https://huggingface.co/fishaudio/drama-3 My FREE resources: https://www.skool.com/ai-automation-society/about?el=roast-room-2&hcategory=youtube-videos&utm_campaign=free-group My Tools💻 FREE MONTH voice to text: https://get.glaido.com/nate Code NATEHERK for 10% off VPS (annual plan): https://www.hostinger.com/vps/claude-code-hosting I walk through my Roast Room in Claude Code, where A…
Thanks to Neon for sponsoring this video. Try Neon here : https://neon.plug.dev/Sw0JRgk Google's TimesFM 3.0 forecasts time series without any training on your data. Hand it the history, and it continues the chart with a range around every day. In this video, I show how it works (patches, tokens, "autocomplete for charts"), what changed from 2.5 to 3.0, and I test both versions on real New York subway ridership and on a simulated business running on Neon. Then I give the exact planning decisio…
This AI shows every shopper a different popup and keeps optimizing for revenue, right before Black Friday. Book a demo and mention Nate Herk for a 30 day free trial: https://hubs.la/Q04yvprG0 #ad
Learn more about LLM Inference here → https://ibm.biz/~CCvvB5TvB Every LLM response starts with much more than a prompt. Grace Ableidinger explains LLM inference, from tokenization and GPU batching to KV cache, prefill, and decode. Learn how AI models turn static weights into real-time responses. 00:00 – Understand What LLM Inference Actually Does 01:15 – See How LLM Inference Engines Serve AI Models 01:43 – Follow an LLM Request from API to Model 02:15 – Turn Prompts Into Tokens with Tok…
📺 Discover how to create rich and interactive content with HTML pages in SharePoint. This demo walks through creating a page with Copilot in SharePoint, uploading existing HTML, editing content, and publishing and sharing your finished page. See how to turn SharePoint list information into a project status update and explore flexible layouts, animated elements, anchor links, and an Excel-connected dashboard that updates on refresh. Learn how to bring existing HTML files into the Site Pages li…
🚨 Everything you’re about to see, I benchmarked using the tool I made. If you want to run these tests yourself or test models on whatever you actually use them for! → https://www.woaibench.ai/ Google's Gemini 4 Carbon might have just received a MASSIVE upgrade, and the latest checkpoint is producing some of the most impressive AI-generated 3D environments we've seen yet! Could Gemini 4 Carbon finally compete with Claude Opus 5.5? Today, we're breaking down the latest Arena leaks, insane Voxel…
Anthropic just announced Claude Duet, a new app where Claude works as a proactive agent. It comes to you first instead of waiting for a prompt. Duet runs on Claude Opus 5.5 on its own computer in the cloud, so it keeps working after you close the app. Each morning it sends you a short brief from your connected apps, and during the day it messages you only when something needs your attention. Before it sends a message, makes a payment, signs a document, or deletes anything, it shows you what it’…
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
AI helped uncover something dangerous in the software behind a huge part of the internet. Government websites went offline as a precaution while defenders raced to respond. Here's how OpenAI and Google's models helped researchers uncover PHP2Shell, why finding the flaws was only the beginning, and what you can do to protect yourself. ______________________________________________ My Links 🔗 ➡️ Twitter: https://x.com/WesRothMoney ➡️ AI Newsletter: https://natural20.beehiiv.com/subscribe Want …
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.com/forwardfuture 👉🏻 Instagram: https://www.instagram.com/matthewberman_ai 👉🏻 Discord: https://discord.gg/u7wTTGWhuJ 👉🏻 Spotify: https://open.spotify.com/show/6dBxDwxtHl1hpqHhfoXmy8 Media/Sponsorship Inquiries ✅ https://bit.ly/44TC45V
In this solo episode, I break down the war on AI agents, from Amazon blocking Meta's Muse to Delta treating outside booking agents as unauthorized automated activity to Yelp charging agents for access. I share how I think about an agent, as a customer with a credit card who reads your site instead of looking at it, and the ladder of trust that runs from research to prepare to pay. Then I walk through each big player's bet, my nine stages for how this war plays out, five moves any business can m…
Every senior engineer's answer is "it depends." Max Kanat-Alexander says what it actually depends on. Code quality always existed for two things: understanding a system and safely changing it. AI is making a new kind of software common: disposable software, with low consequences, few dependents and near-zero cost to replace or maintain. For that software, quality doesn't matter. For durable software it matters more than ever. He also shows how to make systems more disposable and how to make rig…
Sam Altman joins me to break down AI consciousness, recursive self-improvement, and how AI could help discover new medicines and scientific breakthroughs. We also get into the future of AI devices, the pressure of leading OpenAI, and why Sam believes we should aim for a new renaissance. Huge thanks to Sam for joining me! https://x.com/sama Join My Newsletter for Regular AI Updates 👇🏼 https://forwardfuture.com My Links 🔗 👉🏻 X: https://x.com/matthewberman 👉🏻 Forward Future X: https://x.…
DeepSeek just open-sourced DeepInfer, the inference engine behind its own API, along with a detailed write-up of how it serves DeepSeek V4.1 Flash at scale: 155 trillion tokens a day from a single cluster. In this video we follow one request through the system: deepseek-recipe tokenization and prompt-injection protection, a KV cache that DeepSeek keeps for days (96% of input tokens are cache hits), prefill and decode on different machines, expert parallelism across 64+ GPUs, DSpark speculative …
That melted ice cream might not be the freezer’s fault. On episode 56 of Security Intelligence, Claire Nunez and MJ Vaidya join host Patrick Austin to break down three cybersecurity stories and what they mean for defenders. First, attackers hijack refrigeration controllers from Danfoss, Copeland and Dixell at French supermarkets, pushing cases to 50°C. Team82 flagged the flaws and Danfoss shipped a fix, yet thousands of controllers remain exposed online. Is planned downtime the price of sta…
Monzo's in-house agent started as an emoji you could react with in Slack. Now it authors over 10% of all merged PRs across Monzo's tech teams and runs more than 1,800 tasks a day. Felipe and Francesco explain why it worked so fast. Their answer is the "lots of little multipliers" Monzo already had: an MCP gateway into internal tools and knowledge, a sandboxed agent orchestration system built for a regulated bank, and a platform that turns the agent itself into a building block. That building bl…
Matt Pocock skills v1.3 adds three new skills I find really helpful, and where each fits in the AI software development lifecycle. /implement-spec runs your tickets with parallel AI agents, /pr makes AI pull requests fast to review with proof and risk, and /retro learns from agent mistakes to improve your AGENTS.md and guardrails. 👋 𝗛𝗲𝘆, 𝗜’𝗺 𝗘𝗿𝗶𝗰. I spent seven years as a senior software engineer at Amazon and Microsoft. Now I am the founder of bookzero.ai, which I built with Claude …
Learn more about New Frontier AI Models here → https://ibm.biz/~QV9RAqelj A confident AI decision isn't necessarily a trustworthy one. Jeff Crume examines Jev through a cybersecurity lens, from attack surfaces and poisoned inputs to observability and verification. Learn why AI decisions need evidence, auditability, and human oversight. 00:00 – Question Whether AI Confidence Can Be Trusted 00:26 – Understand the Problem with Jev 01:47 – Compare Jev Decisions with LLM Outputs 02:48 – Separa…
THIS JUST IN: You need a better headshot. Create yours now on ChatGPT Images.
Anthropic’s new Public Benefit Fund commits roughly 20 percent of company shares to the public good, and the debate reveals a bigger fight over AI wealth, power, and democracy. NLW breaks down the praise and criticism, plus Anthropic’s proposed AI safety bill and why safety advocates are upset. In the headlines: the OpenCode cyber attack, AI bubble fears, Microsoft pulling a UK data center deal, and New York City’s call to pause the Anthropic IPO. The AI Daily Brief helps you understand the mo…
Banks don't buy AI software until it survives their security and compliance review. Al Chen shares what it took to deploy inside five major banks without a year-long bottleneck. He covers the build pipeline, CVE remediation with agents writing and testing exploits, backup and restore verification, PII leaking through LLM judges, and why zero leakage and data sovereignty are never free. Chapters: - 00:00 Intro - 02:15 What banks check - 04:17 Fixing CVEs with agents - 08:30 Proving backup and r…
🚨 Everything you’re about to see, I benchmarked using the tool I made. If you want to run these tests yourself or test models on whatever you actually use them for! → https://www.woaibench.ai/ HUGE GPT-6.5 LEAK?! OpenAI, Anthropic, Google, MiniMax, and Moonshot AI are making massive moves as we uncover some of the biggest AI leaks and model releases of the week! Join the private community to get WoAI Bench Pro, exclusive AI freebies and discounts, early model access, automation templates, pr…
“You Can See Everything,” the new Nathan Fielder-Elizabeth Holmes movie, is a masterclass in manipulation, Caitlin Roper explains on the latest episode of "Hard Fork." Roper runs film, TV and books at The New York Times. She wrote the first-ever magazine piece about Elizabeth Holmes for Wired back in 2013. Join our discussion on the movie at nytimes.com/hardfork.
Is the agent harness dead? Jediah Katz (SpaceXAI, formerly Cursor’s longest-tenured agent-harness engineer) explains what smarter models mean for coding agents, Grokbot and the future of agent harnesses. He shows why fancy harnesses no longer add coding intelligence: the data flywheel teaches models the skills harnesses used to patch in. Then he argues the harness matters more than ever. It opens new frontiers like Grokbot, and it owns what models can’t absorb: infrastructure, trust and securi…
https://github.com/agentSystemLabs/nebula https://github.com/webdevcody/survive-the-night-fps https://survivethenightgame.com/
Build, Learn, Earn with AI: https://www.skool.com/earlyaidopters/about Scan Your App Free with Lockpick: https://lockpick.app Work With Us: https://www.promptadvisers.com/ Someone got into an old OpenCode system, and the API keys people saved in its Console may have been exposed. One developer woke up to over $400 in charges overnight. Dax Raad, one of the key people building OpenCode, says they need AI agents running 24/7 now, because that's what the attackers seem to have. In this video I b…
Get Your Complete Financial OS at https://startup-ideas-pod.link/brex_SIP In this solo episode, I break down how to make money with Meta Muse by mining Facebook groups, Marketplace, and Instagram to match buyers with sellers. The idea came after my wife and I had a baby and the agency we hired for a night nurse admitted it found candidates in Facebook groups. I demo Muse finding night nurses at $30 an hour in Montreal, show how to give it a phone number through Bland AI, and share the Marketpl…
A new essay from former OpenAI researcher Nick Cammarata is being hailed as a potentially historic AI breakthrough. His argument: neural networks may already be far more understandable than researchers believed, and years of apparent complexity may reflect flaws in how we’ve been looking at them. NLW breaks down the surprisingly simple connection between cat faces and car fronts, why bigger models could be easier to interpret, and what this could mean for detecting AI deception—with a look at w…
Bringing legal expertise and frontier AI together starts with trust. For Thomson Reuters’ Camillo Speroni, that means output lawyers can verify, cite and put their name on. Camillo discusses Thomson Reuters’ work with OpenAI to bring HighQ and Thomson Reuters CoCounsel Legal into ChatGPT Enterprise, connecting the tools lawyers rely on in one place.
Hostinger Black Friday deals: http://hostinger.com/nateherkvps FREE AI OS Template: https://app.aiautomationsociety.ai/optin/ai-os/?utm_source=youtube&utm_medium=organic&utm_content=100-ai-tools&sl=yt_100-ai-tools How I built an AI agency from 0 to over $1 million per year: https://www.skool.com/ai-automation-society-plus/about?el=100-ai-tools&hcategory=youtube-videos&utm_campaign=plus-group My FREE resources: https://www.skool.com/ai-automation-society/about?el=100-ai-tools&hcategory=youtube-v…
DeepSeek calls V4.1 Flash 1017 a “minor update”, but it went up on every benchmark in their table, and it’s now the #1 open model on Arena’s WebDev leaderboard. The bigger story is tokens. DeepSeek has always been cheap per token, but its Flash models were token-hungry, so cheap per token never meant cheap per task. In my reruns, 1017 used about a third fewer thinking tokens at high effort and 45% fewer at max, with the same prompts and settings. 🔗 Links DeepSeek announcement: https://x.com/…
Claim your free $5 on Crusoe Intelligence Foundry today 👉https://console.crusoecloud.com/request-foundry?utm_source=tinahuang&utm_medium=influencer&utm_campaign=mi-su-credits In this video, I walk through how to learn AI with different learning paths and resources. 📚 Get the complete resource document/prompts: https://resource.lonelyoctopus.com/signup/ranking-ai-courses/?utm_source=youtube&utm_medium=video&utm_campaign=ranking_courses&utm_content=ranking_courses 🐙 Free 28-Day AI Sprint R…
Learn more about Loop Engineering here → https://ibm.biz/~LqlzZo42O AI agents need more than one perfectly written prompt. Martin Keen explains loop engineering and how AI agents act, observe, adjust, and repeat toward measurable goals. Learn how context, tools, subagents, and human oversight make agentic loops work. 00:00 – Understand Loop Engineering vs Prompt Engineering 01:18 – Build an AI Agent Loop Around a Goal 02:30 – Define Goals AI Agents Can Actually Verify 03:29 – Engineer Tri…
Deploying AI into real businesses is hard: org charts are messy, trust takes decades, and service firms don't scale like software. Prathik Naidu, founding partner at Long Lake, explains their strategy: acquire established service businesses, then transform them with AI from the inside. He covers how they choose companies and what changes once they own them, drawing on 40 businesses across residential services, architecture, HR and more. Chapters: - 00:00 Intro - 02:15 Why deploying AI into bus…
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambda.ai/papers 📝 Gemini 4 Argon + report are available here: https://deepmind.google/models/model-cards/gemini-4-argon/ https://storage.googleapis.com/deepmind-media/gemini/gemini_4_safety_report.pdf Sources: https://x.com/thebuggeddev/status/2110241991233790456 https://x.com/vib3coded/status/2110136719827747181 https://x.com/EnvolDev/status/2110116583016419760 https://x.com/kickingkeys/status/2110037855452180633 https://x.c…
🚨 Everything you’re about to see, I benchmarked using the tool I made. If you want to run these tests yourself or test models on whatever you actually use them for! → https://www.woaibench.ai/ The AI world just had another INSANE weekend! Anthropic has announced a massive new Public Benefit Fund, committing an initial 5% of its shares toward initiatives that could potentially include direct financial distributions to the public. Could this be the beginning of AI-funded Universal Basic Income …
You’ll want to turn the sound on for this one. Read more here: https://chatgpt.com/features/voice/
Ein kleiner Tipp von uns: Ton an, denn das willst du hören. Hier weiterlesen: https://chatgpt.com/de-DE/features/voice/
Pour cette vidéo, pensez à activez le son pour ne rien manquer. En savoir plus: https://chatgpt.com/fr-FR/features/voice/
Insurance claims are full of handwriting, faxes, duplicates and 8GB PDFs. Denys Linkov of Wisedocs explains why that unstructured data is both the biggest barrier and the biggest opportunity for AI in insurance. He covers cleaning the data, analysis within a single document, and cross-claim insights, along with the benchmark Wisedocs built with Artificial Analysis to test how models handle medical records. Chapters: - 00:00 Intro - 01:43 Insurance's technology transformation - 09:34 How a clai…
A hands-on workshop on evaluating AI applications where mistakes are expensive. Annabell Schäfer and Doneyli De Jesus from Langfuse use a banking assistant to walk through the whole AI engineering loop. They cover tracing with PII masking, building datasets, turning domain-expert comments into failure modes, LLM-as-judge and code evaluators, and CI that blocks a prompt change when scores drop. Chapters: - 00:00 Why evals are hard - 05:01 The AI engineering loop - 12:47 Demo: a banking assistan…
Your agent says it's done. Is it? Nick Nisi and Zack Proser from WorkOS show how to build verification loops that let coding agents check their own work. They cover clear goals, CLI health checks, browser snapshots, LLM judges and evals, so a failure drives the next fix instead of landing on your desk. Includes the live "Midnight Snack Club" demo app and the repo to follow along. Chapters: - 00:00 Why "done" isn't done - 03:17 What it takes to trust agent output - 09:34 Give agents goals they …
🚨 Everything you’re about to see, I benchmarked using the tool I made. If you want to run these tests yourself or test models on whatever you actually use them for! → https://www.woaibench.ai/ DeepSeek is BACK, and its latest model, DeepSeek V4.1 Flash 1017, might just be one of the most impressive open-weight AI models we've seen! In this video, we put DeepSeek's newest Flash model to the test, exploring its insane coding capabilities, 3D world generation, agentic performance, benchmarks, an…
Can RL Make a 4B Model Compete with 27B? My Open RL Environment Experiments (GRPO, LoRA, DGX) In this video I test whether open-source RL environments can make a small model perform closer to a much larger one by training Qwen 3.5 4B (with a LoRA) against Kaggle-style “small data” RL tasks on two DGX Sparks and comparing to Qwen 27B on a held-out exam. I explain what RL environments are (sandbox, tools, checker), why you need a strict checker to avoid reward hacking, and how GRPO works and onl…
Clay's new agent builds your whole outbound system from one prompt, and you can run it from Claude Code or Codex right now. Go to my Instagram, DM me "SCULPT" and I'll send it over. #ad
If your token spend keeps going up and productivity isn't keeping pace, start here. Eugene Cheah lays out four steps for cutting AI inference costs at scale: measure, right-size, buy capacity well and check your processes. Along the way he covers splitting API keys per team, agentic vs. process workloads, open models that match frontier models on specific tasks, classifiers, fine-tuning, the speed-vs-cost tradeoff and "slop grenades." Recorded live at AI Engineer NYC 2026. 🎟️ Join us at the …
The model is the part you can't change. The harness around it is where you can make real gains. Nacho Martinez walks through the full agent stack (agent loop, short- and long-term memory, context engineering, the toolbox and skillbox patterns, skill distillation and continual learning) and applies it to an autonomous agent that flags suspicious bank transactions around the clock. 🎟️ Join us at the next one: AIE CODE Summit → https://aie.codes Speaker Nacho Martinez — LinkedIn: https://www.li…
🚨 Everything you’re about to see, I benchmarked using the tool I made. If you want to run these tests yourself or test models on whatever you actually use them for! → https://www.woaibench.ai/ Claude Fable 5.5 could be our first glimpse of BABY AGI?! Anthropic's mysterious next-generation AI model is showing some absolutely insane capabilities, from generating interactive 3D worlds to recreating complex simulations using just code! Join the private community to get WoAI Bench Pro, exclusive…