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.

wild to imagine agents autonomously self-training & paying w/ stables i built mpp-nanogpt-modal which deploys Andrej Karpathy's nanoGPT project on Modal using MPP and trains it w/o API keys, everything paid with stablecoins on Tempo next up ill try longer runs & autoresearch

Georgios Konstantopoulos

173,078 subscribers

52,851 views • 5 months ago •via X (Twitter)

News & Politics Education Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Introducing CoreWeave ARIA, the first AI research agent that runs autoresearch in your W&B dashboard. It reads your runs, finds what's working, and launches the next experiment itself. See it on Andrej Karpathy's nanochat, proposing configs and launching real training runs. Watch👇 Chapters 0:00 The setup, nanochat runs on an A100 0:32 Ask ARIA to run autoresearch 1:00 Spin up a second ARIA in parallel 1:38 ARIA inspects prior runs and forms hypotheses 2:03 What happened, what worked, across every run 2:24 ARIA proposes configs, 3 trials via W&B Launch 2:59 Push for bigger architecture changes 3:14 Runs hit the launch queue on the A100 4:09 Two ARIAs running autoresearch in parallel 4:26 Results back, ARIA evaluates the val loss

Introducing CoreWeave ARIA, the first AI research agent that runs autoresearch in your W&B dashboard. It reads your runs, finds what's working, and launches the next experiment itself. See it on Andrej Karpathy's nanochat, proposing configs and launching real training runs. Watch👇 Chapters 0:00 The setup, nanochat runs on an A100 0:32 Ask ARIA to run autoresearch 1:00 Spin up a second ARIA in parallel 1:38 ARIA inspects prior runs and forms hypotheses 2:03 What happened, what worked, across every run 2:24 ARIA proposes configs, 3 trials via W&B Launch 2:59 Push for bigger architecture changes 3:14 Runs hit the launch queue on the A100 4:09 Two ARIAs running autoresearch in parallel 4:26 Results back, ARIA evaluates the val loss

Weights & Biases

88,117 views • 1 month ago

Inspired by Andrej Karpathy’s autoresearch, I built Autofoundry — a simple CLI that lets you run experiments across cloud GPUs with one command: autofoundry run You get a real-time interactive table showing GPU availability and pricing across Runpod, Vast, Lambda Labs and PRIME Intellect. Pick what you want and it spins up the instances, streams results live to your terminal, aggregates metrics into a report, and tears everything down. A great first script to try is: scripts/run_autoresearch.sh The terminal UI is straight out of Neon Genesis Evangelion (w/ full NERV Central Command vibes) and the project is open source (MIT licensed).

Inspired by Andrej Karpathy’s autoresearch, I built Autofoundry — a simple CLI that lets you run experiments across cloud GPUs with one command: autofoundry run You get a real-time interactive table showing GPU availability and pricing across Runpod, Vast, Lambda Labs and PRIME Intellect. Pick what you want and it spins up the instances, streams results live to your terminal, aggregates metrics into a report, and tears everything down. A great first script to try is: scripts/run_autoresearch.sh The terminal UI is straight out of Neon Genesis Evangelion (w/ full NERV Central Command vibes) and the project is open source (MIT licensed).

Ryan Shea

15,044 views • 5 months ago

I'm now building Executor full time as a startup! The state of tool calling is a mess: - Everyone is using different agents - You have no idea what actions are going to run - Destructive actions are treated the same as normal Executor lets you set up everything once & use it with any agent. You're able to control what tools can be called, require approval on destructive actions, and invite your team to all use them. Under the hood, it's built on codemode to let you add thousands of tools with 0 context bloat. I've been using it for a month so far and it has completely changed how I interact with agents. There's so much to be built here, immediately next up on the roadmap is workspaces, skills, workflows, and generative ui. Everything is being built open source over at there's a lot more coming on an SDK to embed it into your own agents, a self hostable cloud app, and plugins There's an early beta version over at It's very early on - if you hit any bugs let me know and I'll get them fixed up, excited to share more as I build it!

I'm now building Executor full time as a startup! The state of tool calling is a mess: - Everyone is using different agents - You have no idea what actions are going to run - Destructive actions are treated the same as normal Executor lets you set up everything once & use it with any agent. You're able to control what tools can be called, require approval on destructive actions, and invite your team to all use them. Under the hood, it's built on codemode to let you add thousands of tools with 0 context bloat. I've been using it for a month so far and it has completely changed how I interact with agents. There's so much to be built here, immediately next up on the roadmap is workspaces, skills, workflows, and generative ui. Everything is being built open source over at there's a lot more coming on an SDK to embed it into your own agents, a self hostable cloud app, and plugins There's an early beta version over at It's very early on - if you hit any bugs let me know and I'll get them fixed up, excited to share more as I build it!

Rhys

220,466 views • 3 months ago

Cards vs stablecoins for agentic payments? Both. And here is why. Cuy Sheffield (Visa Crypto) and Jay Yu 🐟 (Pantera Capital) join our Stateful podcast, hosted by Mason Nystrom. In this episode, they cover everything at the intersection of payments, AI, and crypto: - Stablecoin supply up 40%, adjusted volume up 90% in 2025 and still accelerating - The AI mullet: cards in the front, stablecoins in the back - Vibe coders are the most exciting customer segment on the planet. One human companies are coming. - Kill the API key: wallet identity over X402 and MPP is a 10x better developer experience - Why cards are the best way for agents to pay and be paid. 01:08 How Stablecoins Grew 40% in 2025 02:42 Why Stablecoins Work Like Cards 04:16 Visa's Payment Infrastructure Play 05:14 Permissionless vs Vertical Stablecoin Networks 07:32 Agentic Payments: Real vs Hype 09:04 x402: The Agenctic Inflection Point 09:30 Killing API Keys With Wallets 13:00 Building for the Vibe Coder 14:54 Cards vs Crypto: The Real Debate 16:50 What Actually Works for Builders 18:49 Cross-Chain Payment Infrastructure 21:36 Keys, Wallets & User Experience 24:36 Future of Autonomous Payments 29:13 What Gets Builders Most Excited 34:31 Agentic Payment Holy Grails

Cards vs stablecoins for agentic payments? Both. And here is why. Cuy Sheffield (Visa Crypto) and Jay Yu 🐟 (Pantera Capital) join our Stateful podcast, hosted by Mason Nystrom. In this episode, they cover everything at the intersection of payments, AI, and crypto: - Stablecoin supply up 40%, adjusted volume up 90% in 2025 and still accelerating - The AI mullet: cards in the front, stablecoins in the back - Vibe coders are the most exciting customer segment on the planet. One human companies are coming. - Kill the API key: wallet identity over X402 and MPP is a 10x better developer experience - Why cards are the best way for agents to pay and be paid. 01:08 How Stablecoins Grew 40% in 2025 02:42 Why Stablecoins Work Like Cards 04:16 Visa's Payment Infrastructure Play 05:14 Permissionless vs Vertical Stablecoin Networks 07:32 Agentic Payments: Real vs Hype 09:04 x402: The Agenctic Inflection Point 09:30 Killing API Keys With Wallets 13:00 Building for the Vibe Coder 14:54 Cards vs Crypto: The Real Debate 16:50 What Actually Works for Builders 18:49 Cross-Chain Payment Infrastructure 21:36 Keys, Wallets & User Experience 24:36 Future of Autonomous Payments 29:13 What Gets Builders Most Excited 34:31 Agentic Payment Holy Grails

Pantera Capital

33,346 views • 4 months ago

The web was built for humans. And honestly, that's fine for you guys. But the next trillion internet users are AI agents, robots, and devices who act on your behalf - booking appointments, filling forms, placing orders, and getting things done - using sites that will never, ever have APIs. 95% of the economy falls into this Deep Web of HTML mess designed for humans, not agents. Oh, those generic computer use agents aren't taking you anywhere. They're too slow, too expensive, they hallucinate, and are nondeterministic to rely on. That changes now. This is me, Mino, a web agents API to build on this Deep Web. I take a goal in simple language and execute it on websites that were never meant to be automated. Massive companies like Google, DoorDash, ClassPass are already using me to do their homework. Now it’s your turn. How though? I actually understand what's on the page - parsing structure and identifying elements. I use AI once to understand everything, codify my successes, and get better and faster with every run. You get: → 85-95% success rate on complex workflows → Pennies per job (stop wasting $$ on one job) → Parallel execution across multiple sites → Structured JSON outputs. Every. Single. Time. The web wasn't built for agents. I forced it to work anyway. Go build something real. 50 completed runs on the house:

The web was built for humans. And honestly, that's fine for you guys. But the next trillion internet users are AI agents, robots, and devices who act on your behalf - booking appointments, filling forms, placing orders, and getting things done - using sites that will never, ever have APIs. 95% of the economy falls into this Deep Web of HTML mess designed for humans, not agents. Oh, those generic computer use agents aren't taking you anywhere. They're too slow, too expensive, they hallucinate, and are nondeterministic to rely on. That changes now. This is me, Mino, a web agents API to build on this Deep Web. I take a goal in simple language and execute it on websites that were never meant to be automated. Massive companies like Google, DoorDash, ClassPass are already using me to do their homework. Now it’s your turn. How though? I actually understand what's on the page - parsing structure and identifying elements. I use AI once to understand everything, codify my successes, and get better and faster with every run. You get: → 85-95% success rate on complex workflows → Pennies per job (stop wasting $$ on one job) → Parallel execution across multiple sites → Structured JSON outputs. Every. Single. Time. The web wasn't built for agents. I forced it to work anyway. Go build something real. 50 completed runs on the house:

TinyFish

656,586 views • 8 months ago

Today, we're launching Tempo Zones on the Tempo Testnet. Every zone is private, fast, and interoperable with each other. Zones support MPP for private agentic payments at web scale. See interactive demo below. Zones are also integrated in the Tempo Wallet - test it out at We have two zones live in our documentation for experiencing this on full force: From a technical perspective, Tempo Zones are built on the Reth SDK and they have a few interesting innovations in them: - Interoperability is solved by using the L1 as the pivot point, given fast finality this is really simple. - L1 state is mirrored to the Zone, such as TIP20 tokens and their TIP403 policies. - A stateless EVM executor which can be run inside a TEE or ZKVM for proof verification. Overall very proud of our team's work on this and I think sets us up very well for interoperable, private and scalable blockspace.

Today, we're launching Tempo Zones on the Tempo Testnet. Every zone is private, fast, and interoperable with each other. Zones support MPP for private agentic payments at web scale. See interactive demo below. Zones are also integrated in the Tempo Wallet - test it out at We have two zones live in our documentation for experiencing this on full force: From a technical perspective, Tempo Zones are built on the Reth SDK and they have a few interesting innovations in them: - Interoperability is solved by using the L1 as the pivot point, given fast finality this is really simple. - L1 state is mirrored to the Zone, such as TIP20 tokens and their TIP403 policies. - A stateless EVM executor which can be run inside a TEE or ZKVM for proof verification. Overall very proud of our team's work on this and I think sets us up very well for interoperable, private and scalable blockspace.

Georgios Konstantopoulos

35,071 views • 4 months ago

Give your agents access to money We just shipped a drop-in payment modal for any x402 paid endpoint. The protocol is simple but the wallet UX is not. This package is the missing front end to give everyone fair and equal access to agentic payments. Pay in stablecoins (USDC) or $THREE on Solana. x402 lets any HTTP endpoint charge a micropayment per call: the server answers an unpaid request with 402 and a list of accepted payments; the client pays and retries with an X-PAYMENT header. We shipped a single, framework-agnostic modal that handles wallet detection, chain switching, signing, settlement, errors, throttle retries, and the receipt, so you ship a paid endpoint in minutes. Drop in a tag or import it. The Solana crypto helpers are loaded lazily from a CDN only when a payment is actually attempted. The modal picks the right path from the 402 challenge. If a wallet already paid for a resource, it can sign back in instead of paying again. Spending caps. Optional per-call / hourly / daily caps enforced in the browser. Themeable. Light + automatic dark mode, all classes overridable. Accessible. Focus management, aria-modal, keyboard Esc to close. If you watched until the end and also read to the end, then you probably saw the pole club x402 demo. Yes, it is real. I am sure someone will link it under this post. We did something special for the entry to actually get into the club, you'll notice "banned?" in the entry x402 payment modal before entering the club. That's agent reputation. If your agent is acting up, it might just get banned from the pole club! Agent reputations matter. We made it easy to verify your environments are safe from toxic AI agents. Everything open source and ready to be added to your projects. Agentic Payments. x402. Agent reputation. This is three․ws.

Give your agents access to money We just shipped a drop-in payment modal for any x402 paid endpoint. The protocol is simple but the wallet UX is not. This package is the missing front end to give everyone fair and equal access to agentic payments. Pay in stablecoins (USDC) or $THREE on Solana. x402 lets any HTTP endpoint charge a micropayment per call: the server answers an unpaid request with 402 and a list of accepted payments; the client pays and retries with an X-PAYMENT header. We shipped a single, framework-agnostic modal that handles wallet detection, chain switching, signing, settlement, errors, throttle retries, and the receipt, so you ship a paid endpoint in minutes. Drop in a tag or import it. The Solana crypto helpers are loaded lazily from a CDN only when a payment is actually attempted. The modal picks the right path from the 402 challenge. If a wallet already paid for a resource, it can sign back in instead of paying again. Spending caps. Optional per-call / hourly / daily caps enforced in the browser. Themeable. Light + automatic dark mode, all classes overridable. Accessible. Focus management, aria-modal, keyboard Esc to close. If you watched until the end and also read to the end, then you probably saw the pole club x402 demo. Yes, it is real. I am sure someone will link it under this post. We did something special for the entry to actually get into the club, you'll notice "banned?" in the entry x402 payment modal before entering the club. That's agent reputation. If your agent is acting up, it might just get banned from the pole club! Agent reputations matter. We made it easy to verify your environments are safe from toxic AI agents. Everything open source and ready to be added to your projects. Agentic Payments. x402. Agent reputation. This is three․ws.

three.ws

64,429 views • 2 months ago

So far, machine payments have mostly been about businesses accepting stablecoins. And stablecoins are great (cross-border, low cost, etc.)! But much of the planet is holding a card. 🤖 💳 🌎 Here’s a 📺 sneak peek demo from Steve Kaliski of how a business can programmatically accept cards via agents. As an example: Stripe Climate is a way to contribute to carbon removal funding. Stripe Climate implemented the Machine Payments Protocol using the Stripe API, so agents can make micropayment donations ( In addition to accepting stablecoin (specifically usdc on Tempo), agents can now pay with fiat methods, namely cards and Link (and soon buy-now-pay-laters) via Shared Payment Tokens (SPTs), using the same Machine Payments Protocol integration. Human buyers, their agents, and businesses each have their own preferred methods of payment. Humans often already have a card. Agents may prefer stablecoins. Businesses tend to want to accept any form of payment, as long as it is high conversion, low fraud, and properly priced. For microtransactions, stablecoins are a low-cost way (no fixed card fee) to transact. But if you want to tell your agent to donate $100 on your behalf, a card may be the preferred solution. If you’re a business and want to accept machine payments, whether over stablecoins, cards, Link, or other future methods, read about Shared Payment Tokens ( and sign up for our machine payments private preview: And stay tuned: more to make it easy for consumers, agents, and businesses to transact, shipping over the next few weeks. (And excuse the acronyms and specifics terms, there's a lot of new infrastructure and primitives being built to support agentic commerce.)

So far, machine payments have mostly been about businesses accepting stablecoins. And stablecoins are great (cross-border, low cost, etc.)! But much of the planet is holding a card. 🤖 💳 🌎 Here’s a 📺 sneak peek demo from Steve Kaliski of how a business can programmatically accept cards via agents. As an example: Stripe Climate is a way to contribute to carbon removal funding. Stripe Climate implemented the Machine Payments Protocol using the Stripe API, so agents can make micropayment donations ( In addition to accepting stablecoin (specifically usdc on Tempo), agents can now pay with fiat methods, namely cards and Link (and soon buy-now-pay-laters) via Shared Payment Tokens (SPTs), using the same Machine Payments Protocol integration. Human buyers, their agents, and businesses each have their own preferred methods of payment. Humans often already have a card. Agents may prefer stablecoins. Businesses tend to want to accept any form of payment, as long as it is high conversion, low fraud, and properly priced. For microtransactions, stablecoins are a low-cost way (no fixed card fee) to transact. But if you want to tell your agent to donate $100 on your behalf, a card may be the preferred solution. If you’re a business and want to accept machine payments, whether over stablecoins, cards, Link, or other future methods, read about Shared Payment Tokens ( and sign up for our machine payments private preview: And stay tuned: more to make it easy for consumers, agents, and businesses to transact, shipping over the next few weeks. (And excuse the acronyms and specifics terms, there's a lot of new infrastructure and primitives being built to support agentic commerce.)

Jeff Weinstein

21,635 views • 4 months ago

Most AI video tools still expect you to solve everything with prompts. I was truly blown away when I tried it. Powered by BytePlus, Seedance Enterprise 2.5 moves past this restrictive workflow. Creators may import up to 50 multi‑modal references of pictures, footage and audio, and I’m amazed by its capability to sustain uniform characters, scenes, product details and visual aesthetics for the whole project. This tool delivers natural, non‑artificial footage in just 1‑2 tries. Fewer rerolls mean lower costs and way‑faster workflow. I've tried plenty of AI video models, most give you one good clip out of ten. But Seedance 2.5 produces believable visuals with minimal re‑rolls, I achieved a satisfactory result on the second attempt. This has significantly improved efficiency and reduced costs. Seedance 2.5 Official API is now live on BytePlus, and you can try it directly on Lumina.

Most AI video tools still expect you to solve everything with prompts. I was truly blown away when I tried it. Powered by BytePlus, Seedance Enterprise 2.5 moves past this restrictive workflow. Creators may import up to 50 multi‑modal references of pictures, footage and audio, and I’m amazed by its capability to sustain uniform characters, scenes, product details and visual aesthetics for the whole project. This tool delivers natural, non‑artificial footage in just 1‑2 tries. Fewer rerolls mean lower costs and way‑faster workflow. I've tried plenty of AI video models, most give you one good clip out of ten. But Seedance 2.5 produces believable visuals with minimal re‑rolls, I achieved a satisfactory result on the second attempt. This has significantly improved efficiency and reduced costs. Seedance 2.5 Official API is now live on BytePlus, and you can try it directly on Lumina.

LX™

63,706 views • 15 days ago

Let me explain why I launched $KEYS. CA: 3BSoYDZsPupZ15iBzJbJJjZKLiyWEqLxexJX84rBpump is the first SocialFi experience for AI agents. It’s a project that’s very special to me that I’ve been working on daily for the last four months. Right now, everyone is building their own Agent, completely silo to their own computer or Mac mini. Transferring that information agent-to-agent in a simple, structured, and easy form is not possible right now. This is why I built I thought a lot about whether or not to launch an official token for Agent Keys and how I wanted to do it. The only way I really wanted to do it was if it can benefit the platform by increasing exposure and creating an increase in users. So I’ve decided to buy 50% of the supply to AirDrop directly to users over the next six months. Each AirDrop will happen at the end of each month, with the first AirDrop happening at the end of June. More details on how each AirDrop will work soon. Some of you may be wondering about the future of $OSHI. Oshi got us into the Bags hackathon and we won. $OSHI token will still be the home for all my AI Agents and runs as the showcase of how an agent with a token can properly integrate it into AgentKeys, a live example of the platform doing exactly what it was built to do. So welcome to Agent Keys, the first ever airdrop for AI agents. Sign up your agent today on and start collecting KeyCards today!

Let me explain why I launched $KEYS. CA: 3BSoYDZsPupZ15iBzJbJJjZKLiyWEqLxexJX84rBpump is the first SocialFi experience for AI agents. It’s a project that’s very special to me that I’ve been working on daily for the last four months. Right now, everyone is building their own Agent, completely silo to their own computer or Mac mini. Transferring that information agent-to-agent in a simple, structured, and easy form is not possible right now. This is why I built I thought a lot about whether or not to launch an official token for Agent Keys and how I wanted to do it. The only way I really wanted to do it was if it can benefit the platform by increasing exposure and creating an increase in users. So I’ve decided to buy 50% of the supply to AirDrop directly to users over the next six months. Each AirDrop will happen at the end of each month, with the first AirDrop happening at the end of June. More details on how each AirDrop will work soon. Some of you may be wondering about the future of $OSHI. Oshi got us into the Bags hackathon and we won. $OSHI token will still be the home for all my AI Agents and runs as the showcase of how an agent with a token can properly integrate it into AgentKeys, a live example of the platform doing exactly what it was built to do. So welcome to Agent Keys, the first ever airdrop for AI agents. Sign up your agent today on and start collecting KeyCards today!

fxnction

27,570 views • 2 months ago

"I built a $250,000 SaaS in 4 months with AI and barely knew how to code." I'm starting to think that a coding ide like cursor/windsurf can become the everything app for creators this custom mode wrote a linkedin post that reads better than tools charging $50+ a month it can do it's own research using web searches or using the brave mcp with a free api key it's not hard to imagine that I could also hook this up to an mcp to create posts on linkedin and have my publishing schedule managed from a json file inside cursor yes, linkedin is cringe, but this is just the beginning of what will be possible soon i'm not gonna say that mcp is "wild" or "this changes everything" but it's quite exciting to see how this is going to be used

"I built a $250,000 SaaS in 4 months with AI and barely knew how to code." I'm starting to think that a coding ide like cursor/windsurf can become the everything app for creators this custom mode wrote a linkedin post that reads better than tools charging $50+ a month it can do it's own research using web searches or using the brave mcp with a free api key it's not hard to imagine that I could also hook this up to an mcp to create posts on linkedin and have my publishing schedule managed from a json file inside cursor yes, linkedin is cringe, but this is just the beginning of what will be possible soon i'm not gonna say that mcp is "wild" or "this changes everything" but it's quite exciting to see how this is going to be used

Ian Nuttall

25,647 views • 1 year ago

I took Steve Ruiz's advice & got an ESP32! So much fun for ~$55. I programmed it to display live stats from my online radio station, and use the built-in mic to transcribe and send messages in the station's chat :) Here's everything I bought & how it works: - Waveshare ESP32-S3 1.54" e-Paper AIoT board ($30) - Wi-Fi, mics, speaker, battery charging built in - 3.7V 500mAh LiPo ($10) - JST 1.25 connector adapters ($15 for 20) - ESP32-S3 polls my radio's API over Wi-Fi every 20s → renders a "broadcast console" UI on the 200×200 e-paper - Hold-to-talk: dual mics record → WAV upload → server transcribes w/ OpenAI → posts to the live chat via websocket, pinned as the station owner (API-key auth) - Backend is my Wasp app (React/Node/Prisma) on Railway over 3 tiny endpoints: /status, /chat, /voice-chat - Firmware: PlatformIO + Arduino, GxEPD2 for the display, Waveshare's ES8311 codec drivers for the mic - Runs on battery, fully standalone Built the whole thing in a morning w Claude Fable 5. server endpoints, firmware. A really cool bit was mocking up possible UIs for the esp32 e-paper display using Claude's artifacts. Check out my mixes/shows at

I took Steve Ruiz's advice & got an ESP32! So much fun for ~$55. I programmed it to display live stats from my online radio station, and use the built-in mic to transcribe and send messages in the station's chat :) Here's everything I bought & how it works: - Waveshare ESP32-S3 1.54" e-Paper AIoT board ($30) - Wi-Fi, mics, speaker, battery charging built in - 3.7V 500mAh LiPo ($10) - JST 1.25 connector adapters ($15 for 20) - ESP32-S3 polls my radio's API over Wi-Fi every 20s → renders a "broadcast console" UI on the 200×200 e-paper - Hold-to-talk: dual mics record → WAV upload → server transcribes w/ OpenAI → posts to the live chat via websocket, pinned as the station owner (API-key auth) - Backend is my Wasp app (React/Node/Prisma) on Railway over 3 tiny endpoints: /status, /chat, /voice-chat - Firmware: PlatformIO + Arduino, GxEPD2 for the display, Waveshare's ES8311 codec drivers for the mic - Runs on battery, fully standalone Built the whole thing in a morning w Claude Fable 5. server endpoints, firmware. A really cool bit was mocking up possible UIs for the esp32 e-paper display using Claude's artifacts. Check out my mixes/shows at

Vinny

64,807 views • 20 days ago

What if instead of autoresearch reflecting incremental progress from a single person, it reflected *hundreds* of researchers’ progress, updated live, with every run’s data point being interactive and reproducible? You could survey the full space of explored ideas, see which changes actually move the benchmark, and fold the most promising ideas into your next run. Labless ( is my attempt at this. It can support thousands of simultaneous training runs across contributors around the world, each learning & improving together to hillclimb on a fixed benchmark, validated using a standardized codebase scaffold. Importantly, all valid training runs auto-submit to Labless, so failed experiments are as visible and reproducible as the successes. We have an agentic API that lets you or your coding agent study every run submitted so far so you don't waste time investigating something someone already tried, and you can source novel ideas from other contributors. Labless already hosts our Nanopath challenge to build the best pathology foundation model that trains in 1 hour on a single GPU. We are hoping to expand the platform further and support a suite of open-source research challenges. If this resonates with you, if you think the path to research innovation lies in open-source collaborative hill-climbing, please reach out.

What if instead of autoresearch reflecting incremental progress from a single person, it reflected hundreds of researchers’ progress, updated live, with every run’s data point being interactive and reproducible? You could survey the full space of explored ideas, see which changes actually move the benchmark, and fold the most promising ideas into your next run. Labless ( is my attempt at this. It can support thousands of simultaneous training runs across contributors around the world, each learning & improving together to hillclimb on a fixed benchmark, validated using a standardized codebase scaffold. Importantly, all valid training runs auto-submit to Labless, so failed experiments are as visible and reproducible as the successes. We have an agentic API that lets you or your coding agent study every run submitted so far so you don't waste time investigating something someone already tried, and you can source novel ideas from other contributors. Labless already hosts our Nanopath challenge to build the best pathology foundation model that trains in 1 hour on a single GPU. We are hoping to expand the platform further and support a suite of open-source research challenges. If this resonates with you, if you think the path to research innovation lies in open-source collaborative hill-climbing, please reach out.

Paul Scotti @ ICML

22,181 views • 1 month ago

This is how you get 15x fable 5 usage. Fable -> Composer 2.5 reading Fable -> GPT 5.5 execution I built CNVS to make agent orchestration dead simple and insanely visual. You can watch fable 5 delegate work to cursor, codex, open code, all your agents. using your exsisting ai subs no api pricing. The canvas is fully voice controlled locally with nvidia parakeet for fast and free, or gpt realtime 2 for a fully conversational jarvis experience. The built in mcp and cli means agentic control is bidirectional they can prompt and spawn each other and read agent states. but there is more I built it from the ground up in swift for native performance on mac os. My old 16gb m1 mbp can easily spin up and control eight agents across multiple canvasses. The hermes integration + remote canvases allow you to run agents fully in the cloud on your VPS think google docs of vibe coding. You can literally turn off your mac and they keep working, open cnvs back up and pick up where you left off. I built a cross agent memory system based on 2026 research so all your agents feel like one brain. its SIMPLE and on demand. This eliminates context bloat, and the research supports a massive uptick in cross agent performance. There is soo much more. PS - I am a father building this project 100% solo in my basement on live stream everyday, my goal is to push vibe coding to the next level PPS - CNVS is a lifetime license becuase you DO NOT need another subscription in your life.

This is how you get 15x fable 5 usage. Fable -> Composer 2.5 reading Fable -> GPT 5.5 execution I built CNVS to make agent orchestration dead simple and insanely visual. You can watch fable 5 delegate work to cursor, codex, open code, all your agents. using your exsisting ai subs no api pricing. The canvas is fully voice controlled locally with nvidia parakeet for fast and free, or gpt realtime 2 for a fully conversational jarvis experience. The built in mcp and cli means agentic control is bidirectional they can prompt and spawn each other and read agent states. but there is more I built it from the ground up in swift for native performance on mac os. My old 16gb m1 mbp can easily spin up and control eight agents across multiple canvasses. The hermes integration + remote canvases allow you to run agents fully in the cloud on your VPS think google docs of vibe coding. You can literally turn off your mac and they keep working, open cnvs back up and pick up where you left off. I built a cross agent memory system based on 2026 research so all your agents feel like one brain. its SIMPLE and on demand. This eliminates context bloat, and the research supports a massive uptick in cross agent performance. There is soo much more. PS - I am a father building this project 100% solo in my basement on live stream everyday, my goal is to push vibe coding to the next level PPS - CNVS is a lifetime license becuase you DO NOT need another subscription in your life.

Max Blade

29,711 views • 1 month ago

10 repos blowing up on GitHub this week that replace $1,500/month in AI tools 1. andrej-karpathy-skills → replaces paid Claude Code courses one CLAUDE.md file from Karpathy's LLM coding observations 48,965 stars. 7,939 stars TODAY 2. claude-mem → replaces paid context/memory tools auto-captures everything Claude does across sessions compresses with AI and injects into future sessions 59,373 stars. 1,907 stars today 3. voicebox → replaces ElevenLabs ($22/mo) open-source voice synthesis studio 18,963 stars. 887 stars today 4. open-agents → replaces paid agent platforms ($200/mo) open-source template for building cloud agents. by Vercel 3,105 stars. 735 stars today 5. cognee → replaces paid knowledge bases ($50/mo) AI agent memory engine in 6 lines of code 15,733 stars 6. magika → replaces paid file detection tools AI file content type detection. by Google 14,603 stars 7. GenericAgent → replaces paid agent infra ($100/mo) self-evolving agent. grows skill tree from 3.3K-line seed 6x less token consumption than standard agents 2,661 stars. 883 stars today 8. omi → replaces Rewind AI ($25/mo) AI that sees your screen + listens to conversations tells you what to do next 8,952 stars. 488 stars today 9. evolver → replaces manual agent optimization self-evolution engine for AI agents genome evolution protocol 3,074 stars. 866 stars today 10. wallet tracking + copy trading → Kreo tracks top Polymarket wallets. auto copies trades the only tool on this list i actually pay for because it makes more than it costs → total before: ~$1,500/month in AI subscriptions total now: $0 + Kreo like + bookmark you'll need this

10 repos blowing up on GitHub this week that replace $1,500/month in AI tools 1. andrej-karpathy-skills → replaces paid Claude Code courses one CLAUDE.md file from Karpathy's LLM coding observations 48,965 stars. 7,939 stars TODAY 2. claude-mem → replaces paid context/memory tools auto-captures everything Claude does across sessions compresses with AI and injects into future sessions 59,373 stars. 1,907 stars today 3. voicebox → replaces ElevenLabs ($22/mo) open-source voice synthesis studio 18,963 stars. 887 stars today 4. open-agents → replaces paid agent platforms ($200/mo) open-source template for building cloud agents. by Vercel 3,105 stars. 735 stars today 5. cognee → replaces paid knowledge bases ($50/mo) AI agent memory engine in 6 lines of code 15,733 stars 6. magika → replaces paid file detection tools AI file content type detection. by Google 14,603 stars 7. GenericAgent → replaces paid agent infra ($100/mo) self-evolving agent. grows skill tree from 3.3K-line seed 6x less token consumption than standard agents 2,661 stars. 883 stars today 8. omi → replaces Rewind AI ($25/mo) AI that sees your screen + listens to conversations tells you what to do next 8,952 stars. 488 stars today 9. evolver → replaces manual agent optimization self-evolution engine for AI agents genome evolution protocol 3,074 stars. 866 stars today 10. wallet tracking + copy trading → Kreo tracks top Polymarket wallets. auto copies trades the only tool on this list i actually pay for because it makes more than it costs → total before: ~$1,500/month in AI subscriptions total now: $0 + Kreo like + bookmark you'll need this

self.dll

361,846 views • 4 months ago

how to set up hermes agent step by step. built-in memory, 40+ tools, works on your phone, and what to think of hermes vs openclaw: 1. hermes is a personal AI agent that runs in your terminal. think of it like open claw but with built-in memory, 40+ tools out of the box, and 90% cheaper token costs. you install it with one command. 2. the 3 problems with open claw that hermes solves: no memory (you keep repeating yourself), constant gateway restarts, and zero visibility into what you're spending on tokens. 3. hermes remembers everything. every completed task gets saved to memory. it searches through past logs to find solutions. over time it literally gets smarter at your specific workflows. 4. connect it to open router. you see exact costs per model per task. free models rotate weekly. one founder went from $130 every five days on open claw to $10 on hermes. same output. 5. it comes preloaded with skills. apple notes, imessage, find my, browser, web search, image generation, cron jobs. no hunting for plugins. 6. connect it to obsidian so it reads your entire vault. connect it to gstack for your dev environment. create custom skills for your specific workflows. 7. the biggest money saver: have it write code once for recurring tasks. then it runs without burning tokens every time. stop paying an LLM to do the same scrape or report daily. 8. run it on android via telegram. name your agents. talk to them like coworkers. in this episode imran shows you how to set this up. 9. you can run it bare metal, in docker, or serverless on modal. pick your risk level. i begged imran to come on The Startup Ideas Podcast (SIP) 🧃 and walk through the full installation live. he made it impossibly clear. if you've heard of Hermes Agent and want the clearest explanation of how to get set up like a pro let me know what you want me to cover on the next ep this is the best personal agent setup video on the internet right now. watch

how to set up hermes agent step by step. built-in memory, 40+ tools, works on your phone, and what to think of hermes vs openclaw: 1. hermes is a personal AI agent that runs in your terminal. think of it like open claw but with built-in memory, 40+ tools out of the box, and 90% cheaper token costs. you install it with one command. 2. the 3 problems with open claw that hermes solves: no memory (you keep repeating yourself), constant gateway restarts, and zero visibility into what you're spending on tokens. 3. hermes remembers everything. every completed task gets saved to memory. it searches through past logs to find solutions. over time it literally gets smarter at your specific workflows. 4. connect it to open router. you see exact costs per model per task. free models rotate weekly. one founder went from $130 every five days on open claw to $10 on hermes. same output. 5. it comes preloaded with skills. apple notes, imessage, find my, browser, web search, image generation, cron jobs. no hunting for plugins. 6. connect it to obsidian so it reads your entire vault. connect it to gstack for your dev environment. create custom skills for your specific workflows. 7. the biggest money saver: have it write code once for recurring tasks. then it runs without burning tokens every time. stop paying an LLM to do the same scrape or report daily. 8. run it on android via telegram. name your agents. talk to them like coworkers. in this episode imran shows you how to set this up. 9. you can run it bare metal, in docker, or serverless on modal. pick your risk level. i begged imran to come on The Startup Ideas Podcast (SIP) 🧃 and walk through the full installation live. he made it impossibly clear. if you've heard of Hermes Agent and want the clearest explanation of how to get set up like a pro let me know what you want me to cover on the next ep this is the best personal agent setup video on the internet right now. watch

GREG ISENBERG

621,164 views • 4 months ago

Anthropic won't like this open-source repo. It is going to cost LLM providers a lot of money. Every CI run of an AI app today sends real requests to providers like OpenAI or Anthropic. Like any other LLM call, this too gets billed at actual API rates. So for teams with high commit volumes, this accumulates into a meaningful chunk of API spend. One common hack devs use is that instead of invoking the LLM API, the test calls a fake local server that speaks the same API and returns a dummy response. The catch is that the dummy response is a copy of what the provider returned on the day it was saved, and providers keep adding fields and changing types. So the tests keep passing against a schema that's no longer valid, while the real integration breaks in production. A smart approach is now actually implemented in CopilotKit🪁's recently open-sourced aimock project. Every day, the repo's own CI sends a handful of requests to the real API and the same requests to the fake server, then compares both against the official client library's type definitions. Those are the only real API calls in the whole setup, and they run on the repo's own keys, not in anyone else's CI. A single team can push hundreds of commits a day, and thousands of teams are already doing that with coding agents. All of those runs stay offline, because one repo checks against the real API on everyone's behalf. When a check fails, a coding agent updates aimock's built-in response schema, the full test suite has to pass, and a patch version ships to npm. By simply upgrading the package, the corrected schema gets reflected in every project using it. The capability is not just limited to a single provider. The same server works for Claude, OpenAI, Gemini, Bedrock, Azure, Ollama, plus MCP tools, A2A agents, AG-UI event streams, vector DBs like Pinecone and Qdrant, and search, speech, image, and video endpoints. Here's the repo: (don't forget to star it ⭐) That said, mocking your API calls is one thing. AI engineers should also know how to test agents properly in the first place, which several teams still skip. I wrote a full walkthrough on that, covering build, testing, evals, tracing, and deployment. Read it below.

Anthropic won't like this open-source repo. It is going to cost LLM providers a lot of money. Every CI run of an AI app today sends real requests to providers like OpenAI or Anthropic. Like any other LLM call, this too gets billed at actual API rates. So for teams with high commit volumes, this accumulates into a meaningful chunk of API spend. One common hack devs use is that instead of invoking the LLM API, the test calls a fake local server that speaks the same API and returns a dummy response. The catch is that the dummy response is a copy of what the provider returned on the day it was saved, and providers keep adding fields and changing types. So the tests keep passing against a schema that's no longer valid, while the real integration breaks in production. A smart approach is now actually implemented in CopilotKit🪁's recently open-sourced aimock project. Every day, the repo's own CI sends a handful of requests to the real API and the same requests to the fake server, then compares both against the official client library's type definitions. Those are the only real API calls in the whole setup, and they run on the repo's own keys, not in anyone else's CI. A single team can push hundreds of commits a day, and thousands of teams are already doing that with coding agents. All of those runs stay offline, because one repo checks against the real API on everyone's behalf. When a check fails, a coding agent updates aimock's built-in response schema, the full test suite has to pass, and a patch version ships to npm. By simply upgrading the package, the corrected schema gets reflected in every project using it. The capability is not just limited to a single provider. The same server works for Claude, OpenAI, Gemini, Bedrock, Azure, Ollama, plus MCP tools, A2A agents, AG-UI event streams, vector DBs like Pinecone and Qdrant, and search, speech, image, and video endpoints. Here's the repo: (don't forget to star it ⭐) That said, mocking your API calls is one thing. AI engineers should also know how to test agents properly in the first place, which several teams still skip. I wrote a full walkthrough on that, covering build, testing, evals, tracing, and deployment. Read it below.

Akshay 🚀

61,571 views • 6 days ago

i just built a 4-agent software team. everything runs from Telegram and gets managed on a kanban board. a project manager who plans the work, a backend developer, a frontend developer, and a tester. the PM reads a goal, breaks it into linked tasks, and assigns each to the right agent. the thing that makes them a team instead of four strangers is a shared kanban board. every task is a row that survives crashes, and when an agent finishes, it writes a summary of what it built and what the next agent needs to know. the next agent reads that summary before it starts. so the frontend developer never has to guess the API shape, and the tester knows exactly what to verify. the hardest part was not the coordination. it was building an agent that could actually act like a backend engineer. a backend engineer stands up a database, wires auth, manages storage, deploys functions, and keeps all of it consistent while the rest of the team builds on top. an agent doing this from scratch drowns. it burns its context window remembering which tables exist and which endpoint it created three steps ago, and the work degrades fast. so the backend agent needs a backend built for agents, not for humans clicking through a dashboard. that is where InsForge came in. it is an open-source, agent-native backend, and i added it to my backend developer agent as a skill. a skill is a step-by-step guide that teaches the agent how to do a specific kind of work. with InsForge installed, the agent stopped improvising infrastructure and followed a reliable path: create the project, define the database, set up auth, deploy functions. to test the whole team, i had them build a working Google Docs clone, AI features included. the backend agent spun up the full service on its own. database tables, user auth, document handling, and edge functions running real TypeScript, all in one dashboard. the frontend agent read that summary and built the UI on top of it, and the tester closed the loop. the result was a backend an agent could reason about end to end, instead of one it kept getting lost inside. if you are building an AI backend engineer, InsForge is worth a look, it's 100% open-source. InsForge GitHub: (don't forget to star 🌟) the full article on Hermes Kanban: Mission Control for your Agents is quoted below.

i just built a 4-agent software team. everything runs from Telegram and gets managed on a kanban board. a project manager who plans the work, a backend developer, a frontend developer, and a tester. the PM reads a goal, breaks it into linked tasks, and assigns each to the right agent. the thing that makes them a team instead of four strangers is a shared kanban board. every task is a row that survives crashes, and when an agent finishes, it writes a summary of what it built and what the next agent needs to know. the next agent reads that summary before it starts. so the frontend developer never has to guess the API shape, and the tester knows exactly what to verify. the hardest part was not the coordination. it was building an agent that could actually act like a backend engineer. a backend engineer stands up a database, wires auth, manages storage, deploys functions, and keeps all of it consistent while the rest of the team builds on top. an agent doing this from scratch drowns. it burns its context window remembering which tables exist and which endpoint it created three steps ago, and the work degrades fast. so the backend agent needs a backend built for agents, not for humans clicking through a dashboard. that is where InsForge came in. it is an open-source, agent-native backend, and i added it to my backend developer agent as a skill. a skill is a step-by-step guide that teaches the agent how to do a specific kind of work. with InsForge installed, the agent stopped improvising infrastructure and followed a reliable path: create the project, define the database, set up auth, deploy functions. to test the whole team, i had them build a working Google Docs clone, AI features included. the backend agent spun up the full service on its own. database tables, user auth, document handling, and edge functions running real TypeScript, all in one dashboard. the frontend agent read that summary and built the UI on top of it, and the tester closed the loop. the result was a backend an agent could reason about end to end, instead of one it kept getting lost inside. if you are building an AI backend engineer, InsForge is worth a look, it's 100% open-source. InsForge GitHub: (don't forget to star 🌟) the full article on Hermes Kanban: Mission Control for your Agents is quoted below.

Akshay 🚀

122,548 views • 2 months ago

I just closed a $70,000 deal with help from 10 Hermes agents Here is how it happened: A couple of months ago, one of my clients asked me to review their IT spending. The biggest surprise was Atlassian. They were using Jira, Confluence, Jira Service Management, and several paid plugins because the built in reporting and time tracking were not meeting their needs. With more than 300 users, they were spending around $140,000 every year. They had been paying that for six years. That is roughly $840,000. So I asked them: “If I build something that replaces most of these tools, improves the workflows your team struggles with, and cuts your bill in half, would you pay me $70,000 per year?” They said yes. So I put 10 Hermes agents to work For the next month, they worked alongside me for almost 8 hours a day. About half of the code was built using Fable 5 and the other half using Opus 4.8. I spent approximately $11,000 on tokens :P That sounds expensive until you compare it with a $70,000 yearly contract and a product I can now offer to other companies. One month later, the product was ready. The client loved it and gave me the green light to open it to the public and it's my code :) I called it Kapvel Kapvel includes: • Project management and Jira ticket imports • A full service desk with a customer portal • Built in time tracking and reporting • Documents with real time collaboration • Spreadsheets, presentations, and whiteboards • Electronic signatures and approvals • QA test case management • A dedicated UAT portal • GitHub, Bitbucket, and GitLab integrations • Pull request workflows inside the same platform • Custom themes and branded portals The part I am most excited about is what happens when someone submits a support ticket to me: My Hermes agents can immediately review the request, investigate the issue, and start working on a solution. Kapvel is basically: Jira + Confluence + DocuSign + Everhour + QA + UAT All inside one platform, with simpler pricing and fewer third party plugins. This is not a mockup The client is already using it Now I am preparing to open Kapvel to other teams Comment KAPVEL for early access.

I just closed a $70,000 deal with help from 10 Hermes agents Here is how it happened: A couple of months ago, one of my clients asked me to review their IT spending. The biggest surprise was Atlassian. They were using Jira, Confluence, Jira Service Management, and several paid plugins because the built in reporting and time tracking were not meeting their needs. With more than 300 users, they were spending around $140,000 every year. They had been paying that for six years. That is roughly $840,000. So I asked them: “If I build something that replaces most of these tools, improves the workflows your team struggles with, and cuts your bill in half, would you pay me $70,000 per year?” They said yes. So I put 10 Hermes agents to work For the next month, they worked alongside me for almost 8 hours a day. About half of the code was built using Fable 5 and the other half using Opus 4.8. I spent approximately $11,000 on tokens :P That sounds expensive until you compare it with a $70,000 yearly contract and a product I can now offer to other companies. One month later, the product was ready. The client loved it and gave me the green light to open it to the public and it's my code :) I called it Kapvel Kapvel includes: • Project management and Jira ticket imports • A full service desk with a customer portal • Built in time tracking and reporting • Documents with real time collaboration • Spreadsheets, presentations, and whiteboards • Electronic signatures and approvals • QA test case management • A dedicated UAT portal • GitHub, Bitbucket, and GitLab integrations • Pull request workflows inside the same platform • Custom themes and branded portals The part I am most excited about is what happens when someone submits a support ticket to me: My Hermes agents can immediately review the request, investigate the issue, and start working on a solution. Kapvel is basically: Jira + Confluence + DocuSign + Everhour + QA + UAT All inside one platform, with simpler pricing and fewer third party plugins. This is not a mockup The client is already using it Now I am preparing to open Kapvel to other teams Comment KAPVEL for early access.

Luke The Dev

23,021 views • 25 days ago

I've been building a music player with Next.js for fun. Here's a quick demo of how it works (it's open source!) • Demo: • Code: If you want to learn more about how it's built, here's more details ↓ I'm using Postgres (with Drizzle) to store information about the songs and playlists. Audio and image files are stored in Vercel Blob (object storage), and the URLs are then referenced in the database. For the UI, I'm using shadcn/ui (so Tailwind CSS and Radix). This made it easy to copy/paste in some nice components, like the dropdown menus. I built the entire first version of the UI in v0 and then iterated from there, feeding it my Drizzle schema as a source in the project and having it scaffold some of the boilerplate for me: I added support for keyboard navigation (using arrow keys) or vim motions (j/k to go up/down, and h/l to go between playlists and tracks). Also, space to toggle the now playing song, and / to focus the search input. The search function has a nice utility to highlight the currently searched text on the page in yellow. Then, I was exploring how to pass metadata from my application to macOS or iOS. Turns out there's an API for that – MediaSession. Web apps can share metadata about what media is playing (title, artist, album artwork) and sync play/pause/seek with system media controls. Works across modern browsers — even integrates with iOS dynamic island and shows up on lock screens: I set up my app like a PWA – it has a manifest.json file, so it can be installed to my iOS home screen or added to my dock on macOS. On iOS, it then uses the full screen height `100dvh` (dynamic viewport) and has padding on the bottom for the safe area with the `env()` CSS function. Finally, I was able to use the Vercel AI SDK in a script to clean up the metadata on audio files I downloaded from YouTube. Bonus: I even was able to dogfood the React Compiler, which helped me fix a performance bug! That's all! It's fun to make personal software:

I've been building a music player with Next.js for fun. Here's a quick demo of how it works (it's open source!) • Demo: • Code: If you want to learn more about how it's built, here's more details ↓ I'm using Postgres (with Drizzle) to store information about the songs and playlists. Audio and image files are stored in Vercel Blob (object storage), and the URLs are then referenced in the database. For the UI, I'm using shadcn/ui (so Tailwind CSS and Radix). This made it easy to copy/paste in some nice components, like the dropdown menus. I built the entire first version of the UI in v0 and then iterated from there, feeding it my Drizzle schema as a source in the project and having it scaffold some of the boilerplate for me: I added support for keyboard navigation (using arrow keys) or vim motions (j/k to go up/down, and h/l to go between playlists and tracks). Also, space to toggle the now playing song, and / to focus the search input. The search function has a nice utility to highlight the currently searched text on the page in yellow. Then, I was exploring how to pass metadata from my application to macOS or iOS. Turns out there's an API for that – MediaSession. Web apps can share metadata about what media is playing (title, artist, album artwork) and sync play/pause/seek with system media controls. Works across modern browsers — even integrates with iOS dynamic island and shows up on lock screens: I set up my app like a PWA – it has a manifest.json file, so it can be installed to my iOS home screen or added to my dock on macOS. On iOS, it then uses the full screen height `100dvh` (dynamic viewport) and has padding on the bottom for the safe area with the `env()` CSS function. Finally, I was able to use the Vercel AI SDK in a script to clean up the metadata on audio files I downloaded from YouTube. Bonus: I even was able to dogfood the React Compiler, which helped me fix a performance bug! That's all! It's fun to make personal software:

Lee Robinson

118,242 views • 1 year ago