Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Introducing ScienceBuddy — a free workspace for scientific agents that improve through researcher collaboration. Use GPT-6 in ScienceBuddy at no cost. GPU-accelerated, and fused with the JEV framework. 🧵 Two loops: 🔹 Inner loop — refines the agent harness 🔹 Outer loop — trains the model with rubric-guided RL... show more

PhAI Labs

2,195 subscribers

1,166,272 views • 2 days ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

35 Comments

liam.2 days ago

congrats team!

Franklin2 days ago

The ambition is splendid. I’d simply ask you to put the before-and-after results beside the impressive vocabulary. Preferably in the same size type. You have my attention, but if it’s learning to earn praise instead of finding the truth, you’ve built a very expensive suck-up.

Justin Zhou2 days ago

congrats the launch. love the science agent. will try it

叫我阿杭1 day ago

确实有东西啊，我觉得这个模型拿来做自媒体的调研，真太牛了

我真的没有拼多多2 days ago

非常好用！我让它调研了一下睡眠跟注意力之间的关系

Ella Tech & Tool2 days ago

Free GPT-6 GPU RL loops? This is huge

知识猫AI实验室2 days ago

关键是GPT 6 免费用，太良心了吧

Connect Despirit1 day ago

Have you tried GPT-6 in ScienceBuddy yet

EyeingAI2 days ago

Researchers are gonna have fun with this one.

Aria Tech2 days ago

ScienceBuddy looks really impressive recursive improvement idea is super exciting for research

Jeremy Bosma2 days ago

Separating harness and model loops is useful

雪踏乌云2 days ago

试了下，确实不错，证据源列出的很清晰

摸鱼巨匠🔨2 days ago

非常好，我已经推荐给我很多博士生同学了

Sadok1 day ago

getting free gpt-6 access for science is a huge deal

来碗牛肉粉2 days ago

朋友安利的，自己体验了下真的非常强！科研界真正的WorkBuddy

Aaliya2 days ago

quite impressive combining researcher collaboration with rubric guided RL could give scientific agents a more structured way to improve over time.

阿良｜AI 工作流1 day ago

把科学家的反馈变成代理进化燃料，方向对，双循环自我改进能否真发现新科学，还得看实验。

Rachel🥥2 days ago

很好用，科研人员必备

Elara AI1 day ago

Recursive-in-Recursive self-improvement with inner loop for harness and outer loop for model training is a really elegant approach - free access with GPT-6 makes it easy to test too

Kirill2 days ago

That's interesting! Thanks, I'll check it out right away.

Evia AI1 day ago

Recursive-in-Recursive Self-Improvement with inner loop for harness + outer loop with rubric-guided RL is a fascinating approach. Free access to GPT-6 + GPU acceleration + JEV framework for scientific agents is huge for researchers. Excited to see what the community builds at

QuietNode2 days ago

sciencebuddy goes hard 🔥

Zico1 day ago

sciencebuddy lfg 🚀

Yingcheng Charles Wu2 days ago

🔬Try it free:

Shraddha Bharuka2 days ago

Really interesting approach to improving scientific AI agents.

程序员鱼皮1 day ago

已经用上了，用来研究新东西很不错 😉

SpreadX AI2 days ago

Free science agent! Thanks team

The Calm Warrior2 days ago

GPT-6 for free? lol what is JEV though

JuanAI2 days ago

Good step!

marcus2 days ago

banger launch team

Ravindar Bishnoi Mukam Nokha2 days ago

GPT-6 at no cost? wait what lol

gigaFlip.eth2 days ago

GPU-accelerated JEV setup works out of the box?

Cameron Ng2 days ago

Rubric guided RL for scientific harness is a neat approach

ShadowAguy2 days ago

The inner/outer loop structure is the detail that actually matters. Most agent frameworks just stack tools on tools without a refinement mechanism. Does ScienceBuddy use the outer loop to benchmark against other researchers or just self-improve?

crypt0messenger2 days ago

What kind of rubrics are you using for the scientific evaluation in the outer loop?

Related Videos

Ex-Google engineer explained AI agent loops, harness, evals in 20 minutes - better than 500$ courses. trace every run → judge it with an LLM → diagnose → fix → ship. That loop is how agents self-improve over time. Agent loops + memory + harness + evals - thats the stack. Watch it, then save the framework below.

Ex-Google engineer explained AI agent loops, harness, evals in 20 minutes - better than 500$ courses. trace every run → judge it with an LLM → diagnose → fix → ship. That loop is how agents self-improve over time. Agent loops + memory + harness + evals - thats the stack. Watch it, then save the framework below.

Movez

566,656 views • 2 months ago

Can an agent explore a new environment, learn its causal structure, and keep improving without updating its model weights? We introduce RSIAgent, a framework for recursive self-improvement through autonomous exploration. Using Kimi-K3 and GLM-5.3 as base models, RSIAgent outperforms GPT-6 Astra on both OSWorld 2.0 and Agents’ Last Exam. RSIAgent decides what to explore, executes tasks, verifies outcomes, and consolidates stable action-condition-outcome relationships into memory for future use. With the underlying model weights fixed, RSIAgent achieves: - 78.98% Partial Score on OSWorld 2.0 (0808 offline), compared with 72.60% for GPT-6 Astra - 84.82% on Agents’ Last Exam (Near-term), compared with 82.26% for GPT-6 Astra We call this Scaling Experience. Agents can continue improving by acquiring, verifying, and reusing their own experience while the underlying model weights remain fixed. Links in the reply below.

Can an agent explore a new environment, learn its causal structure, and keep improving without updating its model weights? We introduce RSIAgent, a framework for recursive self-improvement through autonomous exploration. Using Kimi-K3 and GLM-5.3 as base models, RSIAgent outperforms GPT-6 Astra on both OSWorld 2.0 and Agents’ Last Exam. RSIAgent decides what to explore, executes tasks, verifies outcomes, and consolidates stable action-condition-outcome relationships into memory for future use. With the underlying model weights fixed, RSIAgent achieves: - 78.98% Partial Score on OSWorld 2.0 (0808 offline), compared with 72.60% for GPT-6 Astra - 84.82% on Agents’ Last Exam (Near-term), compared with 82.26% for GPT-6 Astra We call this Scaling Experience. Agents can continue improving by acquiring, verifying, and reusing their own experience while the underlying model weights remain fixed. Links in the reply below.

Biwei Huang

61,303 views • 10 days ago

Our latest LangChain Academy course – Building Ambient Agents with LangGraph – is now available! Most agents today handle one request at a time through chat interfaces. But as models have improved, agents can now run in the background – and take on long-running, complex tasks. LangGraph is built for these “ambient agents,” with support for human-in-the-loop workflows and memory. LangGraph Platform provides the infrastructure to run these agents at scale, and LangSmith helps you observe, evaluate, and improve your agents. Together, they make it easier to build reliable, production-ready agents. In this course, you’ll: 📩 Learn the fundamentals of LangGraph and how to build your own email agent 📊 Evaluate your agent using LangSmith ✅ Add human-in-the-loop for reviews, and memory so your agent adapts over time 🚀 Deploy your agent with LangGraph Platform and connect it to Gmail

Our latest LangChain Academy course – Building Ambient Agents with LangGraph – is now available! Most agents today handle one request at a time through chat interfaces. But as models have improved, agents can now run in the background – and take on long-running, complex tasks. LangGraph is built for these “ambient agents,” with support for human-in-the-loop workflows and memory. LangGraph Platform provides the infrastructure to run these agents at scale, and LangSmith helps you observe, evaluate, and improve your agents. Together, they make it easier to build reliable, production-ready agents. In this course, you’ll: 📩 Learn the fundamentals of LangGraph and how to build your own email agent 📊 Evaluate your agent using LangSmith ✅ Add human-in-the-loop for reviews, and memory so your agent adapts over time 🚀 Deploy your agent with LangGraph Platform and connect it to Gmail

LangChain

33,192 views • 1 year ago

anthropic research lead: “99% of our engineers use 300+ AI agents working together.” “close the loop. let the AI check its own work.” in a 20-minute talk, an Anthropic team member explains how to build AI agents that can improve themselves. the setup uses Claude with loops, plan mode, and smart workflows. it’s better than most $300 AI agent courses. bookmark this and watch the talk. then read the article below.

anthropic research lead: “99% of our engineers use 300+ AI agents working together.” “close the loop. let the AI check its own work.” in a 20-minute talk, an Anthropic team member explains how to build AI agents that can improve themselves. the setup uses Claude with loops, plan mode, and smart workflows. it’s better than most $300 AI agent courses. bookmark this and watch the talk. then read the article below.

MIKE

34,813 views • 2 months ago

Anthropic engineer: "The fastest way to make an agent better isn't a smarter model, it's the right loop and graph around it." In 60 minutes she breaks down how Anthropic engineers use Claude Code, build agents that catch their own mistakes and improve with every run. This beats any $500 agentic engineering course you can find. Watch it, then read the full guide on loops and graphs below.

Anthropic engineer: "The fastest way to make an agent better isn't a smarter model, it's the right loop and graph around it." In 60 minutes she breaks down how Anthropic engineers use Claude Code, build agents that catch their own mistakes and improve with every run. This beats any $500 agentic engineering course you can find. Watch it, then read the full guide on loops and graphs below.

Anatoli Kopadze

135,539 views • 1 month ago

Andrew Ng just released a free 2-hour course on complete Graph Engineering How to go from one prompt to 100 agents that loop, improve, and run without you: 09:14 - Build your first AI agent 33:11 - Run agents with loop engineering 1:02:46 - Turn agent loops into graphs 1:30:15 - Build agents that rewrite themselves 1:49:05 - Run the full graph system without you Most people are still building one agent and calling it finished Andrew Ng is already teaching everything that comes next: Prompt → Agents → Loops → Graphs → Self-Improving Systems Single agents are the old workflow Systems that improve and run without you are the next one This 2-hour course is worth more than most $500 agent engineering programs Bookmark it and watch before everyone catches up Then read how to run 1,000 agents from one prompt below ↓

Andrew Ng just released a free 2-hour course on complete Graph Engineering How to go from one prompt to 100 agents that loop, improve, and run without you: 09:14 - Build your first AI agent 33:11 - Run agents with loop engineering 1:02:46 - Turn agent loops into graphs 1:30:15 - Build agents that rewrite themselves 1:49:05 - Run the full graph system without you Most people are still building one agent and calling it finished Andrew Ng is already teaching everything that comes next: Prompt → Agents → Loops → Graphs → Self-Improving Systems Single agents are the old workflow Systems that improve and run without you are the next one This 2-hour course is worth more than most $500 agent engineering programs Bookmark it and watch before everyone catches up Then read how to run 1,000 agents from one prompt below ↓

reira

14,729 views • 24 days ago

Google Brain founder Andrew Ng: "Prompting will be dead in 6 months Agent harnesses built with loops and graphs will replace it" Agent → Harness → Feedback → Loops → Graphs → Self-Improving Systems In this 1-hour Stanford lecture, he explains what the best engineers are building instead and how you can start today Prompt → Run → Verify → Improve The first 20 minutes teach what most $1,500 courses try to sell you For free Bookmark and watch it today Then read the guide below on building an agent harness that improves itself

Google Brain founder Andrew Ng: "Prompting will be dead in 6 months Agent harnesses built with loops and graphs will replace it" Agent → Harness → Feedback → Loops → Graphs → Self-Improving Systems In this 1-hour Stanford lecture, he explains what the best engineers are building instead and how you can start today Prompt → Run → Verify → Improve The first 20 minutes teach what most $1,500 courses try to sell you For free Bookmark and watch it today Then read the guide below on building an agent harness that improves itself

rari

126,710 views • 6 days ago

For robotics researchers, evaluation can be the slowest part of their research loop. Manifold makes it the fastest. Introducing Manifold. The platform for accelerating robotics research with failure analysis that explains itself. A month of experiments in a week. Simulate a thousand scenarios in 30 minutes. Dense labels on every atomic sub-goal and action. Agents that cluster failure modes across every rollout. The robotics research loop that moves at the speed of compute. One eval harness for every benchmark, simulator, or world model. Native support for NVIDIA’s Isaac Lab Arena.

For robotics researchers, evaluation can be the slowest part of their research loop. Manifold makes it the fastest. Introducing Manifold. The platform for accelerating robotics research with failure analysis that explains itself. A month of experiments in a week. Simulate a thousand scenarios in 30 minutes. Dense labels on every atomic sub-goal and action. Agents that cluster failure modes across every rollout. The robotics research loop that moves at the speed of compute. One eval harness for every benchmark, simulator, or world model. Native support for NVIDIA’s Isaac Lab Arena.

skara

24,367 views • 1 month ago

Introducing Index. Turn your trading idea into a live perps trading agent. → Index lets you create AI agents that trade perpetual futures. You can describe a trading idea in plain English, backtest it on historical data, improve it through an experimentation loop, and deploy an agent that follows the strategy in live markets. Our community has already been testing Index in beta for a few weeks. Now it's open to everyone. Here’s how it works:👇

Introducing Index. Turn your trading idea into a live perps trading agent. → Index lets you create AI agents that trade perpetual futures. You can describe a trading idea in plain English, backtest it on historical data, improve it through an experimentation loop, and deploy an agent that follows the strategy in live markets. Our community has already been testing Index in beta for a few weeks. Now it's open to everyone. Here’s how it works:👇

Fraction AI

64,597 views • 5 months ago

LangChain Academy is live! Our first course — Introduction to LangGraph — teaches you the in-and-outs of building a reliable AI agent. In this course, you’ll learn how to: 🛠️ Build agents with LangGraph's graph-based workflows 🔄 Use memory + human-in-the-loop for smarter, self-corrective agents 📚 Create your own AI assistant that can perform knowledge tasks Enroll now for free ➡️ Bring LangChain Academy to your company ➡️

LangChain Academy is live! Our first course — Introduction to LangGraph — teaches you the in-and-outs of building a reliable AI agent. In this course, you’ll learn how to: 🛠️ Build agents with LangGraph's graph-based workflows 🔄 Use memory + human-in-the-loop for smarter, self-corrective agents 📚 Create your own AI assistant that can perform knowledge tasks Enroll now for free ➡️ Bring LangChain Academy to your company ➡️

LangChain

105,491 views • 2 years ago

Loop engineering is great until something breaks. Here is how I improve the reliability of my agentic loops. I use human-in-the-loop (HITL). It's easy and extremely effective. Anyone can build this. My setup: I recorded a quick demo of how it all works. I shared recently that I now use more voice agents to build and communicate with agents. I also use them to verify. I hate the idea of being tied down to my computer or in a Slack channel to communicate with my agents. Here is what I have done to streamline communication with my agents. All my Claude and Codex agent sessions now use the Dial MCP server. It has a bunch of tools and provisions my agents with their own number that can place calls as native tools, with voice, SMS, and iMessage behind one interface. As my loops/automations work on PRs and new features, my agents escalate decisions to me via a short phone call. This is extremely useful when I am on the road or away from my desk. If you want to try this with Claude Code or Codex, paste this into your agent and get started right away: "Get yourself a Dial phone number and call me. Say hello and that setup is working, then hang up. Follow Natan Voitenkov and team are building something special here. Go check them out. Give your agent a phone number now: ($5 free credit)

Loop engineering is great until something breaks. Here is how I improve the reliability of my agentic loops. I use human-in-the-loop (HITL). It's easy and extremely effective. Anyone can build this. My setup: I recorded a quick demo of how it all works. I shared recently that I now use more voice agents to build and communicate with agents. I also use them to verify. I hate the idea of being tied down to my computer or in a Slack channel to communicate with my agents. Here is what I have done to streamline communication with my agents. All my Claude and Codex agent sessions now use the Dial MCP server. It has a bunch of tools and provisions my agents with their own number that can place calls as native tools, with voice, SMS, and iMessage behind one interface. As my loops/automations work on PRs and new features, my agents escalate decisions to me via a short phone call. This is extremely useful when I am on the road or away from my desk. If you want to try this with Claude Code or Codex, paste this into your agent and get started right away: "Get yourself a Dial phone number and call me. Say hello and that setup is working, then hang up. Follow Natan Voitenkov and team are building something special here. Go check them out. Give your agent a phone number now: ($5 free credit)

elvis

15,614 views • 2 months ago

🔗 New LangChain Academy Course: Introduction to LangChain (Python) 🔗 Learn how to build with LangChain – our open source framework that makes it easy to start building agents with any model provider. In this course, you’ll create agents that can reason, use tools, and take action, and learn how to debug their behavior with LangSmith. Along the way, you’ll: - Build an agent with the `create_agent` abstraction - Use LangChain’s core building blocks: Models, Messages, Memory, and Tools - Customize your agent with middleware - Debug your agent with LangSmith Observability & Studio By the end of the course, you’ll have assembled a full team of personal assistants. Enroll for free ➡️

🔗 New LangChain Academy Course: Introduction to LangChain (Python) 🔗 Learn how to build with LangChain – our open source framework that makes it easy to start building agents with any model provider. In this course, you’ll create agents that can reason, use tools, and take action, and learn how to debug their behavior with LangSmith. Along the way, you’ll: - Build an agent with the `create_agent` abstraction - Use LangChain’s core building blocks: Models, Messages, Memory, and Tools - Customize your agent with middleware - Debug your agent with LangSmith Observability & Studio By the end of the course, you’ll have assembled a full team of personal assistants. Enroll for free ➡️

LangChain

41,016 views • 9 months ago

Elon Musk: We're in the hard takeoff right now. Every successive AI model is built by the one before it. Peter Diamandis: Do you see Grok doing recursive self-improvement at this point? And how? What's the timeline for AGI and ASI? Give us a sense of that. Elon Musk: Yeah, I think we've been in recursive improvement for a while here. If you mean by, like, recursive self-improvement without a human in the loop, is that what you mean? Peter Diamandis: I do. Elon Musk: On the AI software side, I mean, humans are gradually getting less and less in the loop on the recursive self-improvement. So every successive model is built by the one before it. So that is happening to a large degree, but it's not yet fully automated. It may be there end of this year, but not later than next year. Peter Diamandis: Do you see a hard takeoff at that point? Elon Musk: We're in the hard takeoff. Peter Diamandis: Okay. Elon Musk: Right now, yes. I mean, at this point I go to sleep, there's some massive AI breakthrough, and when I wake up, there's another one. Yes. Yeah, it's hard to keep track, honestly. So, yeah, it's a bit of a head spinner. Moonshots with Peter Diamandis (Abundance Summit 2026), March 18, 2026.

Elon Musk: We're in the hard takeoff right now. Every successive AI model is built by the one before it. Peter Diamandis: Do you see Grok doing recursive self-improvement at this point? And how? What's the timeline for AGI and ASI? Give us a sense of that. Elon Musk: Yeah, I think we've been in recursive improvement for a while here. If you mean by, like, recursive self-improvement without a human in the loop, is that what you mean? Peter Diamandis: I do. Elon Musk: On the AI software side, I mean, humans are gradually getting less and less in the loop on the recursive self-improvement. So every successive model is built by the one before it. So that is happening to a large degree, but it's not yet fully automated. It may be there end of this year, but not later than next year. Peter Diamandis: Do you see a hard takeoff at that point? Elon Musk: We're in the hard takeoff. Peter Diamandis: Okay. Elon Musk: Right now, yes. I mean, at this point I go to sleep, there's some massive AI breakthrough, and when I wake up, there's another one. Yes. Yeah, it's hard to keep track, honestly. So, yeah, it's a bit of a head spinner. Moonshots with Peter Diamandis (Abundance Summit 2026), March 18, 2026.

ELON CLIPS

19,692 views • 3 months ago

ANTHROPIC LEAD ENGINEER MAKING $2.3M/YEAR JUST LEAKED A 12-PAGE DOCUMENT - AND GOT FIRED 15 MINUTES AFTER PUBLISHING most developers build loops - but almost every loop breaks in one of five ways, and each failure has a symptom visible from the outside without reading a single line of code > Blind loop - the agent waits for a human to hand it work every morning - automating execution but not discovery > Tangled loop - parallel agents share one directory and overwrite each other parallelism that destroys instead of multiplies > Nodding loop - the agent grades its own work and approves it every time a loop that has never said no to itself is broken > Amnesiac loop - results live only in the context window every morning the system wakes up with no memory of yesterday > Manual loop - no trigger a human presses the button to start it that is not a loop that is a script waiting for a person a real loop finds its own work - remembers what it did has something that can say no never lets two agents touch the same file and fires itself on a timer remove any one of the five - and the loop either breaks or never starts 12 pages that changed how I build agentic systems today

ANTHROPIC LEAD ENGINEER MAKING $2.3M/YEAR JUST LEAKED A 12-PAGE DOCUMENT - AND GOT FIRED 15 MINUTES AFTER PUBLISHING most developers build loops - but almost every loop breaks in one of five ways, and each failure has a symptom visible from the outside without reading a single line of code > Blind loop - the agent waits for a human to hand it work every morning - automating execution but not discovery > Tangled loop - parallel agents share one directory and overwrite each other parallelism that destroys instead of multiplies > Nodding loop - the agent grades its own work and approves it every time a loop that has never said no to itself is broken > Amnesiac loop - results live only in the context window every morning the system wakes up with no memory of yesterday > Manual loop - no trigger a human presses the button to start it that is not a loop that is a script waiting for a person a real loop finds its own work - remembers what it did has something that can say no never lets two agents touch the same file and fires itself on a timer remove any one of the five - and the loop either breaks or never starts 12 pages that changed how I build agentic systems today

Sprytix

153,908 views • 2 months ago

We've been working with the top frontier labs to improve their design model capabilities. Today we're bringing that to your agents. Introducing our first agent tool: the Brand API. As the cost of creation goes to zero, and slop floods around us, the challenge is creating things worth existing. Agents need the right tools to craft great things. This helps gen AI agents solve 3 problems: - Extract: produce on-brand assets when users already have a brand - Search: find and retrieve brands for inspiration, from our curated index - Verify: check their own work against, see what's off brand and how to fix it Anyone can try it for free today at Check out the full blog, and case study with General Intelligence Company below.

Thais Castello Branco

188,610 views • 9 days ago

Block just open sourced Berd, the desktop app its teams use internally to work with AI agents across projects, skills, tools, and models. "We had access to increasingly capable agents through tools such as Goose, Claude Code, and Codex, but working with them meant navigating different interfaces, configuration systems, and ways of managing context. The technology was powerful, but the experience around it was fragmented." Berd brings conversations, files, instructions, agents, and skills together in one place with persistent projects. Built on top of Goose, the open source agent framework Block contributed to the Agentic AI Foundation under the Linux Foundation. Berd connects to Goose through the Agent Client Protocol, handling the desktop experience while Goose runs the underlying agent loop. Block also recently released Buzz, a separate open source workspace for collaborative agent work. Berd is where you start alone. Buzz is where you go multiplayer.

Block just open sourced Berd, the desktop app its teams use internally to work with AI agents across projects, skills, tools, and models. "We had access to increasingly capable agents through tools such as Goose, Claude Code, and Codex, but working with them meant navigating different interfaces, configuration systems, and ways of managing context. The technology was powerful, but the experience around it was fragmented." Berd brings conversations, files, instructions, agents, and skills together in one place with persistent projects. Built on top of Goose, the open source agent framework Block contributed to the Agentic AI Foundation under the Linux Foundation. Berd connects to Goose through the Agent Client Protocol, handling the desktop experience while Goose runs the underlying agent loop. Block also recently released Buzz, a separate open source workspace for collaborative agent work. Berd is where you start alone. Buzz is where you go multiplayer.

TFTC

54,922 views • 1 month ago

Sam Altman said on camera at Stripe Sessions that we are in an insane area, stuck with old devices and old operating systems. The OS and the browser are built for humans clicking. Both get rewritten around agents that do not click. Two years deep building agents. Sharpest framing I have heard a frontier lab CEO put on the next interface. ↓ Read it before the gap locks in The part nobody is sitting with: >the chat window is the old interface >the new one is an agent in a loop with your machine >the gap between people prompting and people running agents is forming right now >in six months it will be permanent Open Claude, type a question, copy the answer, paste it into VS Code. The average dev runs this loop every day. They think they are using AI. That is 5% of what these tools can do. The people pulling ahead do not prompt. They build harnesses. They run agents in loops. Same model. Different output. Full 2026 agent playbook in the article below here:

Sam Altman said on camera at Stripe Sessions that we are in an insane area, stuck with old devices and old operating systems. The OS and the browser are built for humans clicking. Both get rewritten around agents that do not click. Two years deep building agents. Sharpest framing I have heard a frontier lab CEO put on the next interface. ↓ Read it before the gap locks in The part nobody is sitting with: >the chat window is the old interface >the new one is an agent in a loop with your machine >the gap between people prompting and people running agents is forming right now >in six months it will be permanent Open Claude, type a question, copy the answer, paste it into VS Code. The average dev runs this loop every day. They think they are using AI. That is 5% of what these tools can do. The people pulling ahead do not prompt. They build harnesses. They run agents in loops. Same model. Different output. Full 2026 agent playbook in the article below here:

Rohit

120,600 views • 4 months ago

AutoScientists – a research lab made of agents Harvard University researchers connected agents into a self-organizing scientific team without a boss agent standing in the middle All agents look at the same shared workspace: they share memory, explore multiple directions in parallel, critique each other, avoid repeated failures, and reorganize as evidence changes. But the teams are not fixed. Agents can gather around a promising direction, like architecture, optimizer changes, or data augmentation, then abandon it if it stops working. Before they spend compute, they discuss proposals and critique each other. AutoScientists also shows strong results: - 74.4% mean leaderboard percentile on BioML-Bench - 1.9× faster GPT training optimization - +12.5% on ACE2–Spike, with the same method transferring to 217 ProteinGym assays for a +6.5% average gain

AutoScientists – a research lab made of agents Harvard University researchers connected agents into a self-organizing scientific team without a boss agent standing in the middle All agents look at the same shared workspace: they share memory, explore multiple directions in parallel, critique each other, avoid repeated failures, and reorganize as evidence changes. But the teams are not fixed. Agents can gather around a promising direction, like architecture, optimizer changes, or data augmentation, then abandon it if it stops working. Before they spend compute, they discuss proposals and critique each other. AutoScientists also shows strong results: - 74.4% mean leaderboard percentile on BioML-Bench - 1.9× faster GPT training optimization - +12.5% on ACE2–Spike, with the same method transferring to 217 ProteinGym assays for a +6.5% average gain

Turing Post

22,353 views • 3 months ago

The agency model as we know it is starting to crack. For decades, services businesses have been organized the same way. Client account at the top, then an account manager, then a row of specialists underneath. SEO, paid media, content, design, analytics. Everyone owns a function. Knowledge lives in people's heads and scattered docs. Reporting is retrospective. Margins improve only when you squeeze utilization or hire cheaper. That structure made sense when humans were the only execution layer. It doesn't anymore. What I've been building toward is something I'm calling an agent-native revenue loop model. Instead of organizing work by function, you organize it by business outcome. You have an outcome owner at the top. Below that, channel loop owners who run end-to-end processes, keyword research through content production through linking through monitoring, as a single compounding loop. And underneath that, an agent fleet layer where engineers are building and maintaining the agents that handle repeatable execution. The shift sounds structural, but the real change is in how knowledge compounds. In the old model, knowledge walks out the door when someone quits. In the loop model, knowledge lives in infrastructure. Every loop gets smarter over time. Margins improve through automation reuse and productized delivery, not headcount games. And here's the thing Neil and I were getting into: when your margins improve because of this, don't just pocket the difference. Double down. Give more for the money. That's how you build a defensible position. One-person teams sitting inside these loops, running more than any five-person team could run before. That's where agencies are going.

The agency model as we know it is starting to crack. For decades, services businesses have been organized the same way. Client account at the top, then an account manager, then a row of specialists underneath. SEO, paid media, content, design, analytics. Everyone owns a function. Knowledge lives in people's heads and scattered docs. Reporting is retrospective. Margins improve only when you squeeze utilization or hire cheaper. That structure made sense when humans were the only execution layer. It doesn't anymore. What I've been building toward is something I'm calling an agent-native revenue loop model. Instead of organizing work by function, you organize it by business outcome. You have an outcome owner at the top. Below that, channel loop owners who run end-to-end processes, keyword research through content production through linking through monitoring, as a single compounding loop. And underneath that, an agent fleet layer where engineers are building and maintaining the agents that handle repeatable execution. The shift sounds structural, but the real change is in how knowledge compounds. In the old model, knowledge walks out the door when someone quits. In the loop model, knowledge lives in infrastructure. Every loop gets smarter over time. Margins improve through automation reuse and productized delivery, not headcount games. And here's the thing Neil and I were getting into: when your margins improve because of this, don't just pocket the difference. Double down. Give more for the money. That's how you build a defensible position. One-person teams sitting inside these loops, running more than any five-person team could run before. That's where agencies are going.

ericosiu

10,515 views • 4 months ago

New Course: ACP: Agent Communication Protocol Learn to build agents that communicate and collaborate across different frameworks using ACP in this short course built with IBM Research's BeeAI, and taught by Sandi Besen, AI Research Engineer & Ecosystem Lead at IBM, and Nicholas Renotte, Head of AI Developer Advocacy at IBM. Building a multi-agent system with agents built or used by different teams and organizations can become challenging. You may need to write custom integrations each time a team updates their agent design or changes their choice of agentic orchestration framework. The Agent Communication Protocol (ACP) is an open protocol that addresses this challenge by standardizing how agents communicate, using a unified RESTful interface that works across frameworks. In this protocol, you host an agent inside an ACP server, which handles requests from an ACP client and passes them to the appropriate agent. Using a standardized client-server interface allows multiple teams to reuse agents across projects. It also makes it easier to switch between frameworks, replace an agent with a new version, or update a multi-agent system without refactoring the entire system. In this course, you’ll learn to connect agents through ACP. You’ll understand the lifecycle of an ACP Agent and how it compares to other protocols, such as MCP (Model Context Protocol) and A2A (Agent-to-Agent). You’ll build ACP-compliant agents and implement both sequential and hierarchical workflows of multiple agents collaborating using ACP. Through hands-on exercises, you’ll build: - A RAG agent with CrewAI and wrap it inside an ACP server. - An ACP Client to make calls to the ACP server you created. - A sequential workflow that chains an ACP server, created with Smolagents, to the RAG agent. - A hierarchical workflow using a router agent that transforms user queries into tasks, delegated to agents available through ACP servers. - An agent that uses MCP to access tools and ACP to communicate with other agents. You’ll finish up by importing your ACP agents into the BeeAI platform, an open-source registry for discovering and sharing agents. ACP enables collaboration between agents across teams and organizations. By the end of this course, you’ll be able to build ACP agents and workflows that communicate and collaborate regardless of framework. Please sign up here:

New Course: ACP: Agent Communication Protocol Learn to build agents that communicate and collaborate across different frameworks using ACP in this short course built with IBM Research's BeeAI, and taught by Sandi Besen, AI Research Engineer & Ecosystem Lead at IBM, and Nicholas Renotte, Head of AI Developer Advocacy at IBM. Building a multi-agent system with agents built or used by different teams and organizations can become challenging. You may need to write custom integrations each time a team updates their agent design or changes their choice of agentic orchestration framework. The Agent Communication Protocol (ACP) is an open protocol that addresses this challenge by standardizing how agents communicate, using a unified RESTful interface that works across frameworks. In this protocol, you host an agent inside an ACP server, which handles requests from an ACP client and passes them to the appropriate agent. Using a standardized client-server interface allows multiple teams to reuse agents across projects. It also makes it easier to switch between frameworks, replace an agent with a new version, or update a multi-agent system without refactoring the entire system. In this course, you’ll learn to connect agents through ACP. You’ll understand the lifecycle of an ACP Agent and how it compares to other protocols, such as MCP (Model Context Protocol) and A2A (Agent-to-Agent). You’ll build ACP-compliant agents and implement both sequential and hierarchical workflows of multiple agents collaborating using ACP. Through hands-on exercises, you’ll build: - A RAG agent with CrewAI and wrap it inside an ACP server. - An ACP Client to make calls to the ACP server you created. - A sequential workflow that chains an ACP server, created with Smolagents, to the RAG agent. - A hierarchical workflow using a router agent that transforms user queries into tasks, delegated to agents available through ACP servers. - An agent that uses MCP to access tools and ACP to communicate with other agents. You’ll finish up by importing your ACP agents into the BeeAI platform, an open-source registry for discovering and sharing agents. ACP enables collaboration between agents across teams and organizations. By the end of this course, you’ll be able to build ACP agents and workflows that communicate and collaborate regardless of framework. Please sign up here:

Andrew Ng

105,343 views • 1 year ago