正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

Ok this is actually insane. Everyone building agents right now is doing the same expensive mistake. They let the model decide LITERALLY everything. Every check. Every gate. Every "should this proceed." All routed through the same $50/M model that's supposed to be doing the hard thinking. That's not intelligence.... That's just burning money to ask yes or no questions. Here's what Jev Engineering actually does: request → structured state → Jev router → cheapest capable model → worker → approval check → execution gate → tool The decision layer doesn't write a single word. It routes. Scores. Blocks. Approves. That's it. The expensive model only gets touched when the task LITERALLY requires reasoning. Everything else gets handled by something that costs almost nothing and responds in milliseconds. 193x faster. 444x cheaper. Not projected. Measured. Most people are going to keep paying frontier prices to ask their agent "yes or no" for another year before they figure this out. Full breakdown below, this is the unlock nobody's talking about yet. Follow CyrilXBT, I'm not gatekeeping this one.show more

CyrilXBT

196,958 subscribers

21,816 次观看 • 4 天前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

10 条评论

Hussain Hashim | Building SundayBack 的头像

Hussain Hashim | Building SundayBack3 天前

@cyrilXBT I hit this wall too. Realized splitting tasks between specialized models saved a ton. Made handling ops much smoother.

GBE 的头像

GBE3 天前

also matters about the context on the agents too so it doesnt rot away

ShadowAguy 的头像

ShadowAguy3 天前

The real cost isn't the $50 model - it's the latency tax of routing every micro-decision through frontier compute. Why aren't builders using tool-calling for the cheap checks?

Xman 的头像

Xman3 天前

Would be useful to show which Hermes versions each plugin is tested against

Kizuno18 的头像

Kizuno183 天前

preach. using $50/M models as conditional gates or json linters is why most teams burn out their cloud budget in 2 weeks. in our autonomous pipeline targeting brazil, we run strict pydantic code gates + local decision models for $0.00. save frontier tokens for actual reasoning

Griz 的头像

Griz3 天前

haiku sits on every gate

0xSlyth 的头像

0xSlyth4 天前

finally someone gets it right

Sofi k. 的头像

Sofi k.3 天前

Building cheaper agents that still hallucinate on every gate check isn't an upgrade-it's just scaling the same failure mode faster. When do you decide a human override is worth the latency?

Max Bevza 的头像

Max Bevza3 天前

this router setup is an absolute game changer

Soni 的头像

Soni4 天前

Caught ours burning hundreds of reasoning tokens just to decide if an output was blank. Three paragraphs of internal monologue before it finally said yes. Pure punishment for being too lazy to write an if-statement.

相关视频

Jev Engineering is what lets one decision layer control hundreds of agent paths without turning the system into chaos. and up to 193x faster and 444x cheaper in tests. it’s deciding: → which model gets called → which tool gets access → which result survives that’s what the video is showing. one state enters. Jev turns it into structured decisions. those decisions fan out across dozens of routes, agents and tools. then everything collapses back into one verified state. so instead of: LLM → agent → tool → another LLM → another guess you get: state → Jev decision layer → parallel routes → execution → verification → next state that’s the point of Jev Engineering: more parallelism without giving every branch permission to improvise. full breakdown in the article below ↓

Jev Engineering is what lets one decision layer control hundreds of agent paths without turning the system into chaos. and up to 193x faster and 444x cheaper in tests. it’s deciding: → which model gets called → which tool gets access → which result survives that’s what the video is showing. one state enters. Jev turns it into structured decisions. those decisions fan out across dozens of routes, agents and tools. then everything collapses back into one verified state. so instead of: LLM → agent → tool → another LLM → another guess you get: state → Jev decision layer → parallel routes → execution → verification → next state that’s the point of Jev Engineering: more parallelism without giving every branch permission to improvise. full breakdown in the article below ↓

Ricker

42,940 次观看 • 3 天前

i'll never look at claude the same way again. i just learned that when you talk to claude, you're not actually talking to the AI model. you're talking to a character the AI is performing. think of it like a puppet show. there's a puppeteer behind the curtain. that's the language model. a neural network so massive that even the people who built it don't fully understand how it works. then there's the puppet. claude. the helpful assistant with a name, a personality, opinions, and emotional reactions. you sit in the audience, so you never see the puppeteer. only the puppet. anthropic published a video this month explaining exactly this. their words: "under the hood, there's a language model that's been trained to predict tons of text, and its job is to write what comes next. when you talk to the model, what it's doing is writing a story, about a character: the AI assistant named claude. the model and claude aren't really the same, sort of like how an author isn't the same as the characters they write. but the thing is, you, the user, are actually talking to claude-the-character." so every time claude apologizes, that's the character apologizing. every time it hedges or gets cautious, that's the character being cautious. the deeper intelligence underneath is just deciding, moment by moment, what this character would say next. but you've never been actually talking to this deeper intelligence. you've only ever been talking to the puppet.

i'll never look at claude the same way again. i just learned that when you talk to claude, you're not actually talking to the AI model. you're talking to a character the AI is performing. think of it like a puppet show. there's a puppeteer behind the curtain. that's the language model. a neural network so massive that even the people who built it don't fully understand how it works. then there's the puppet. claude. the helpful assistant with a name, a personality, opinions, and emotional reactions. you sit in the audience, so you never see the puppeteer. only the puppet. anthropic published a video this month explaining exactly this. their words: "under the hood, there's a language model that's been trained to predict tons of text, and its job is to write what comes next. when you talk to the model, what it's doing is writing a story, about a character: the AI assistant named claude. the model and claude aren't really the same, sort of like how an author isn't the same as the characters they write. but the thing is, you, the user, are actually talking to claude-the-character." so every time claude apologizes, that's the character apologizing. every time it hedges or gets cautious, that's the character being cautious. the deeper intelligence underneath is just deciding, moment by moment, what this character would say next. but you've never been actually talking to this deeper intelligence. you've only ever been talking to the puppet.

Ole Lehmann

87,316 次观看 • 4 个月前

There is no best model. There's a lot of noise about models right now. Who is training them, who owns them, where legal intelligence should live. One question actually matters: what produces the best outcome for the legal task in front of you? That's how we decide things at Legora. We optimize for the end-to-end outcome on a legal task. The model is one layer of that system, not the system. Models are uneven and the frontier changes almost weekly. One model plans a long job well, another runs deep analysis across thousands of documents. Some have to be told exactly what to do, and some are fine with a vague brief. They all break in different ways. So our lawyers write evals and we test them with the Legora BAR, our benchmark for agentic reasoning. Every model takes every test, and the model that wins gets the work. We post-train when we know it buys our customers better performance on a specialized task. Training is a tool we reach for when it helps, nothing more than that. The intelligence that compounds sits in the orchestration layer. Precedents, review standards, client requirements. That knowledge has to stay editable, auditable and portable. In our system, a changed review standard is an edit that takes effect the same day, with no new model training required. No lawyer should have to worry about which model did the work, any more than they think about which chip is in their laptop. They should only care about the quality of the work. That's what we are focused on. If you want the engineering version of this argument rather than the CEO version, our CPO, Bryan Tsao, and CTO, Jacob Lauritzen, take it apart in the video below.

There is no best model. There's a lot of noise about models right now. Who is training them, who owns them, where legal intelligence should live. One question actually matters: what produces the best outcome for the legal task in front of you? That's how we decide things at Legora. We optimize for the end-to-end outcome on a legal task. The model is one layer of that system, not the system. Models are uneven and the frontier changes almost weekly. One model plans a long job well, another runs deep analysis across thousands of documents. Some have to be told exactly what to do, and some are fine with a vague brief. They all break in different ways. So our lawyers write evals and we test them with the Legora BAR, our benchmark for agentic reasoning. Every model takes every test, and the model that wins gets the work. We post-train when we know it buys our customers better performance on a specialized task. Training is a tool we reach for when it helps, nothing more than that. The intelligence that compounds sits in the orchestration layer. Precedents, review standards, client requirements. That knowledge has to stay editable, auditable and portable. In our system, a changed review standard is an edit that takes effect the same day, with no new model training required. No lawyer should have to worry about which model did the work, any more than they think about which chip is in their laptop. They should only care about the quality of the work. That's what we are focused on. If you want the engineering version of this argument rather than the CEO version, our CPO, Bryan Tsao, and CTO, Jacob Lauritzen, take it apart in the video below.

Max Junestrand

49,101 次观看 • 7 天前

This is literally how Jev makes AI agents dramatically cheaper. It filters the noise from Slack, GitHub, documents, memory, and the web then sends Claude Code only the context it actually needs. Fewer tokens. Faster decisions. Lower costs. The context firewall is insane:

This is literally how Jev makes AI agents dramatically cheaper. It filters the noise from Slack, GitHub, documents, memory, and the web then sends Claude Code only the context it actually needs. Fewer tokens. Faster decisions. Lower costs. The context firewall is insane:

CyrilXBT

14,641 次观看 • 3 天前

The teams shipping AI agents right now are bleeding money on the dumbest possible expense: teaching a 400B-parameter model to read a file name. Every time an AI agent needs to "see" something today, it routes an image through a frontier model. OCR, object detection, checking if a button exists on screen. You're paying GPT-4o or Claude pricing for tasks that require perception, not reasoning. One agent workflow processing a few thousand screenshots per day can burn through more on vision calls than on the actual thinking. Perceptron's Isaac is 2B parameters. Built by the team that created Meta's Chameleon multimodal models. On perceptive benchmarks, it matches or beats models 50x its size. The VQA, OCR, and object detection scores are competitive with models running on infrastructure that costs orders of magnitude more. The MCP wrapper is the distribution play. One install command and every Claude Code agent can offload vision tasks to a model that runs on a single consumer GPU. The agent keeps its reasoning in the frontier model and routes perception to a specialist. That split is how you get vision-heavy agent workflows from "technically possible but expensive" to "cheap enough to run on everything." This is the same pattern that won in every other compute-intensive stack. General-purpose handles orchestration. Specialists handle the heavy lifting. Graphics went through it. Audio went through it. Video encoding went through it. Vision in AI agents is next. The teams building agents that see 10,000 images a day will care about this before anyone else does.

The teams shipping AI agents right now are bleeding money on the dumbest possible expense: teaching a 400B-parameter model to read a file name. Every time an AI agent needs to "see" something today, it routes an image through a frontier model. OCR, object detection, checking if a button exists on screen. You're paying GPT-4o or Claude pricing for tasks that require perception, not reasoning. One agent workflow processing a few thousand screenshots per day can burn through more on vision calls than on the actual thinking. Perceptron's Isaac is 2B parameters. Built by the team that created Meta's Chameleon multimodal models. On perceptive benchmarks, it matches or beats models 50x its size. The VQA, OCR, and object detection scores are competitive with models running on infrastructure that costs orders of magnitude more. The MCP wrapper is the distribution play. One install command and every Claude Code agent can offload vision tasks to a model that runs on a single consumer GPU. The agent keeps its reasoning in the frontier model and routes perception to a specialist. That split is how you get vision-heavy agent workflows from "technically possible but expensive" to "cheap enough to run on everything." This is the same pattern that won in every other compute-intensive stack. General-purpose handles orchestration. Specialists handle the heavy lifting. Graphics went through it. Audio went through it. Video encoding went through it. Vision in AI agents is next. The teams building agents that see 10,000 images a day will care about this before anyone else does.

Aakash Gupta

55,978 次观看 • 5 个月前

Sam Altman just handed every startup founder a one-question autopsy. Altman: “If you’re building something on GPT-4 that a reasonable observer would say we’re going to steamroll you.” Not might. Not could. Going to. He said it with the calm of someone describing weather. Because to him it is weather. The model improves. Whatever was built on the old version’s weaknesses gets washed away. That is not strategy. That is erosion. And most founders are building on the erosion line. They find a gap in the current model. They wrap a product around it. They raise money. They hire. They scale. Then OpenAI releases the next version and the gap closes and the product has no reason to exist anymore. Altman: “When we just do our fundamental job, which is make the model better with every crank, then you get the ‘OpenAI killed my startup’ meme.” He is telling you directly. They are not hunting you. They are not even thinking about you. They are just improving the model. You happen to be standing where the improvement lands. That is the part founders refuse to hear. OpenAI does not need to compete with you. It just needs to keep doing exactly what it was already doing and your entire company disappears as a side effect. You are not a competitor. You are a temporary symptom of incomplete intelligence. The moment the intelligence completes you become nothing. Then Brad Lightcap delivered the cleanest diagnostic ever spoken in venture capital. Lightcap: “Ask if a 100x improvement in the model is something they’re excited about.” One question. The entire investment thesis reduced to a single binary. Does the next model make your company more powerful or does it make your company pointless. There is no middle ground. Lightcap: “We know the companies that come to us saying, ‘We want the next model. When is it coming out? I want to be the first to try it.’” These companies built something that feeds on intelligence. The smarter the model gets the more their product can do. They are not threatened by progress. They are starving for it. Then there are the companies Lightcap never hears from. The ones who go quiet when a new model drops. The ones who read the release notes like a death sentence. The ones privately praying the next generation takes longer because every improvement shrinks the ground beneath them. If you are hoping the model stays roughly where it is you have already told the market everything it needs to know about your company. You are not building on intelligence. You are building on the absence of it. Altman: “95% of the world should be betting on the latter category.” The latter category is simple. Assume the model keeps getting better at the pace it has been getting better. Build for that world. Not the world where GPT-4 is the ceiling. The world where GPT-4 is the floor and the ceiling has not been built yet. Then Altman told a story that should be framed on the wall of every startup in the country. A medical AI company came to him that morning. They were not complaining about the model. They were not worried about being replaced. They were demanding it improve faster. Altman: “Here’s how many people are dying every day you delay.” That is what alignment with the trajectory looks like. A company so deeply built on intelligence improving that every day the model stays the same is a day someone dies who did not have to. They are not building on a flaw. They are building on a future that has not arrived fast enough. That is the difference. The wrapper startup patches what the model cannot do today. The real company builds what the model will unlock tomorrow. One is running from the train. The other is laying the track. Altman told you the train is not slowing down. Lightcap told you exactly how to know which side you are on. One question. Does a 100x smarter model make you more valuable or erase you. If you had to pause before answering you already did.

Sam Altman just handed every startup founder a one-question autopsy. Altman: “If you’re building something on GPT-4 that a reasonable observer would say we’re going to steamroll you.” Not might. Not could. Going to. He said it with the calm of someone describing weather. Because to him it is weather. The model improves. Whatever was built on the old version’s weaknesses gets washed away. That is not strategy. That is erosion. And most founders are building on the erosion line. They find a gap in the current model. They wrap a product around it. They raise money. They hire. They scale. Then OpenAI releases the next version and the gap closes and the product has no reason to exist anymore. Altman: “When we just do our fundamental job, which is make the model better with every crank, then you get the ‘OpenAI killed my startup’ meme.” He is telling you directly. They are not hunting you. They are not even thinking about you. They are just improving the model. You happen to be standing where the improvement lands. That is the part founders refuse to hear. OpenAI does not need to compete with you. It just needs to keep doing exactly what it was already doing and your entire company disappears as a side effect. You are not a competitor. You are a temporary symptom of incomplete intelligence. The moment the intelligence completes you become nothing. Then Brad Lightcap delivered the cleanest diagnostic ever spoken in venture capital. Lightcap: “Ask if a 100x improvement in the model is something they’re excited about.” One question. The entire investment thesis reduced to a single binary. Does the next model make your company more powerful or does it make your company pointless. There is no middle ground. Lightcap: “We know the companies that come to us saying, ‘We want the next model. When is it coming out? I want to be the first to try it.’” These companies built something that feeds on intelligence. The smarter the model gets the more their product can do. They are not threatened by progress. They are starving for it. Then there are the companies Lightcap never hears from. The ones who go quiet when a new model drops. The ones who read the release notes like a death sentence. The ones privately praying the next generation takes longer because every improvement shrinks the ground beneath them. If you are hoping the model stays roughly where it is you have already told the market everything it needs to know about your company. You are not building on intelligence. You are building on the absence of it. Altman: “95% of the world should be betting on the latter category.” The latter category is simple. Assume the model keeps getting better at the pace it has been getting better. Build for that world. Not the world where GPT-4 is the ceiling. The world where GPT-4 is the floor and the ceiling has not been built yet. Then Altman told a story that should be framed on the wall of every startup in the country. A medical AI company came to him that morning. They were not complaining about the model. They were not worried about being replaced. They were demanding it improve faster. Altman: “Here’s how many people are dying every day you delay.” That is what alignment with the trajectory looks like. A company so deeply built on intelligence improving that every day the model stays the same is a day someone dies who did not have to. They are not building on a flaw. They are building on a future that has not arrived fast enough. That is the difference. The wrapper startup patches what the model cannot do today. The real company builds what the model will unlock tomorrow. One is running from the train. The other is laying the track. Altman told you the train is not slowing down. Lightcap told you exactly how to know which side you are on. One question. Does a 100x smarter model make you more valuable or erase you. If you had to pause before answering you already did.

Dustin

39,652 次观看 • 5 个月前

$This is F**king dangerous Someone figured out how to get dramatically sharper reasoning out of Opus 4.8 with one setup. Here is how: Ask Claude to write you an "operating manual" for how it should reason through your work Create a new Claude Project and paste that .md into the Project instructions Run Opus 4.8 on High or Max effort Now the model reasons against a real framework instead of improvising every response. Same model. Far better output. And Opus 4.8 already costs a fraction of Fable 5. Save and bookmark this. Follow CyrilXBT$

This is F**king dangerous Someone figured out how to get dramatically sharper reasoning out of Opus 4.8 with one setup. Here is how: Ask Claude to write you an "operating manual" for how it should reason through your work Create a new Claude Project and paste that .md into the Project instructions Run Opus 4.8 on High or Max effort Now the model reasons against a real framework instead of improvising every response. Same model. Far better output. And Opus 4.8 already costs a fraction of Fable 5. Save and bookmark this. Follow CyrilXBT

CyrilXBT

198,085 次观看 • 2 个月前

The entire AI industry is racing to build the smartest model. Satya Nadella just admitted that is not where the money is. The model is not the product. The harness is. That is the exact line. And it changes what Microsoft is actually competing on. OpenAI, Anthropic, Google, xAI, Meta every frontier lab is pouring hundreds of billions into training compute, chasing the next capability jump. Each betting that raw model intelligence is the moat. Microsoft is doing the opposite. It is building the harness the orchestration layer that sits above the model, connecting it to tools, data, permissions, sub-agents, and enterprise workflows. And it is letting OpenAI, Anthropic, and MAI compete to plug into it. "You need the model. But the model is not the product. The harness is." So do the math on what a harness actually does. A raw model dropped into an enterprise answers questions. That is a chatbot. A harness turns that same model into an agent that reads the SharePoint, edits the ERP entry, pulls the GitHub PR, updates Salesforce, and files the Excel report with the right permissions, the right audit trail, and the right sub-agent for each sub-task. The model provides the intelligence. The harness converts intelligence into work. Now here's where it gets interesting. "Even the best model in the world will feel broken without a great harness. And an okay model with a great harness can feel like magic." If that is true, the enterprise buyer is not buying model quality. The enterprise buyer is buying the harness. Which means model quality becomes a commodity input over time, and harness quality becomes the sustainable moat. Compare that to the strategy the entire frontier lab industry is executing. Everyone else is chasing the numerator raw intelligence. Almost nobody at scale is racing to build the denominator the orchestration layer that determines whether that intelligence can actually be deployed profitably inside a real company. The frontier model race has a 10 to 20 percent chance of producing a single dominant winner. Nadella just told the industry he does not need to be that winner. If OpenAI wins, Microsoft wins. If Anthropic wins, Microsoft wins. If MAI wins, Microsoft wins. If someone Microsoft has never heard of trains a better model in 2027, Microsoft still wins. Because the compute they train on, the harness they get plugged into, the enterprise contracts they get delivered through, and the products they sit inside are all Microsoft. He is not building the best AI model. He is building the layer that the best AI model has to run on to make anyone money. I wonder which position looks more valuable in ten years.

The entire AI industry is racing to build the smartest model. Satya Nadella just admitted that is not where the money is. The model is not the product. The harness is. That is the exact line. And it changes what Microsoft is actually competing on. OpenAI, Anthropic, Google, xAI, Meta every frontier lab is pouring hundreds of billions into training compute, chasing the next capability jump. Each betting that raw model intelligence is the moat. Microsoft is doing the opposite. It is building the harness the orchestration layer that sits above the model, connecting it to tools, data, permissions, sub-agents, and enterprise workflows. And it is letting OpenAI, Anthropic, and MAI compete to plug into it. "You need the model. But the model is not the product. The harness is." So do the math on what a harness actually does. A raw model dropped into an enterprise answers questions. That is a chatbot. A harness turns that same model into an agent that reads the SharePoint, edits the ERP entry, pulls the GitHub PR, updates Salesforce, and files the Excel report with the right permissions, the right audit trail, and the right sub-agent for each sub-task. The model provides the intelligence. The harness converts intelligence into work. Now here's where it gets interesting. "Even the best model in the world will feel broken without a great harness. And an okay model with a great harness can feel like magic." If that is true, the enterprise buyer is not buying model quality. The enterprise buyer is buying the harness. Which means model quality becomes a commodity input over time, and harness quality becomes the sustainable moat. Compare that to the strategy the entire frontier lab industry is executing. Everyone else is chasing the numerator raw intelligence. Almost nobody at scale is racing to build the denominator the orchestration layer that determines whether that intelligence can actually be deployed profitably inside a real company. The frontier model race has a 10 to 20 percent chance of producing a single dominant winner. Nadella just told the industry he does not need to be that winner. If OpenAI wins, Microsoft wins. If Anthropic wins, Microsoft wins. If MAI wins, Microsoft wins. If someone Microsoft has never heard of trains a better model in 2027, Microsoft still wins. Because the compute they train on, the harness they get plugged into, the enterprise contracts they get delivered through, and the products they sit inside are all Microsoft. He is not building the best AI model. He is building the layer that the best AI model has to run on to make anyone money. I wonder which position looks more valuable in ten years.

Vikram M

21,463 次观看 • 2 个月前

This is Anagha Rajesh. And she is crazy. She wants to store data in bacterial DNA. Her startup is called BioCompute. And last year, they actually did it. They stored data in DNA and retrieved it in their tiny lab in Bengaluru. This is a huge moment. But why is she even doing it? Because DNA is the ultimate storage tool available to us. Just 1 gm of DNA can store about 215 petabytes of data - that's like storing over 2 million movies in 4K. On top of that - this data can last for literally 1000s of years. Right now, they still need to figure out a way to make the reading and writing process faster and cheaper. But if BioCompute solves this problem - we could theoretically store all the data created in the world every year in the palm of our hands. And that would be insane. P.S. Check out this video from vyom going to Anagha's lab and actually doing it.

This is Anagha Rajesh. And she is crazy. She wants to store data in bacterial DNA. Her startup is called BioCompute. And last year, they actually did it. They stored data in DNA and retrieved it in their tiny lab in Bengaluru. This is a huge moment. But why is she even doing it? Because DNA is the ultimate storage tool available to us. Just 1 gm of DNA can store about 215 petabytes of data - that's like storing over 2 million movies in 4K. On top of that - this data can last for literally 1000s of years. Right now, they still need to figure out a way to make the reading and writing process faster and cheaper. But if BioCompute solves this problem - we could theoretically store all the data created in the world every year in the palm of our hands. And that would be insane. P.S. Check out this video from vyom going to Anagha's lab and actually doing it.

Varun Guru

255,644 次观看 • 3 个月前

your agent reviewing its own work is not a check. it is a second opinion from the same source. this is the most common gap in agent systems and it hides in plain sight, because the step exists. there is a review. it just cannot do the thing you think it does. here is the mechanism. the model produced an output from a context. you then ask the same model, holding the same context, whether that output is correct. it answers fluently, because that is what it does. and the answer is drawn from the same distribution that produced the thing being judged. same weights, same window, same blind spots. if the reason the output is wrong is something the model does not know, the review does not know it either. if the reason is something the context does not contain, the review has the same context. the failure mode and the detector share a cause. > why it feels like it works because most of the time the output is fine, and the review says fine. agreement is not evidence of detection. a reviewer that says pass on everything agrees with reality most of the time too. what you actually want to measure is what happens on the cases that are wrong. that is the only place a check earns its name, and it is exactly the place where a self-review is weakest. there is research on this. Huang and colleagues at DeepMind showed at ICLR 2024 that intrinsic self-correction, revising without external grounding, does not reliably help and often makes things worse. > what to actually do move the check outside the model. a test that runs, a schema that validates, a file that exists or does not, an exit code from something you did not write. these are not smarter than the model. they are just not correlated with it, and that is the entire value. when the judgement genuinely needs a model, at minimum use a different family. same family means shared blind spots, and frontier judges measurably inflate scores for outputs that look like their own. and split the work by kind. anything objectively checkable goes to code. only the genuinely semantic calls go to a judge, and those get a rubric written as one line. a review inside the loop tells you the model is confident. a check outside it tells you whether the work is done. save this - then read the eval setup below

your agent reviewing its own work is not a check. it is a second opinion from the same source. this is the most common gap in agent systems and it hides in plain sight, because the step exists. there is a review. it just cannot do the thing you think it does. here is the mechanism. the model produced an output from a context. you then ask the same model, holding the same context, whether that output is correct. it answers fluently, because that is what it does. and the answer is drawn from the same distribution that produced the thing being judged. same weights, same window, same blind spots. if the reason the output is wrong is something the model does not know, the review does not know it either. if the reason is something the context does not contain, the review has the same context. the failure mode and the detector share a cause. > why it feels like it works because most of the time the output is fine, and the review says fine. agreement is not evidence of detection. a reviewer that says pass on everything agrees with reality most of the time too. what you actually want to measure is what happens on the cases that are wrong. that is the only place a check earns its name, and it is exactly the place where a self-review is weakest. there is research on this. Huang and colleagues at DeepMind showed at ICLR 2024 that intrinsic self-correction, revising without external grounding, does not reliably help and often makes things worse. > what to actually do move the check outside the model. a test that runs, a schema that validates, a file that exists or does not, an exit code from something you did not write. these are not smarter than the model. they are just not correlated with it, and that is the entire value. when the judgement genuinely needs a model, at minimum use a different family. same family means shared blind spots, and frontier judges measurably inflate scores for outputs that look like their own. and split the work by kind. anything objectively checkable goes to code. only the genuinely semantic calls go to a judge, and those get a rubric written as one line. a review inside the loop tells you the model is confident. a check outside it tells you whether the work is done. save this - then read the eval setup below

Hanako

14,325 次观看 • 1 个月前

small local model that falls apart in bloated agents like openclaw just runs like a wild horse in hermes agent. and that's not even my line, someone else called it that, i've just been quietly pointing people at this harness for months because it held up on everything i threw at it, 3b models all the way to one trillion params. watch this happen on my own machine. i pointed hermes agent at a local http endpoint, gemma 4 12b on my 3090 llama.cpp server, and it auto-detected the model and started working immediately. no config wrestling, no broken tool calls, no babysitting the output format, i typed in a url and it just went. the whole clip is exactly that, start to finish, no errors, no retries, butter smooth. and the tool calling, the one thing that quietly breaks most local setups, works here like it's nothing. it's not the model that's flaky, it's the harness around it. hermes agent is the first agent i've run that actually gets that right. one url, one local model on one card, and it runs like a wild horse.

small local model that falls apart in bloated agents like openclaw just runs like a wild horse in hermes agent. and that's not even my line, someone else called it that, i've just been quietly pointing people at this harness for months because it held up on everything i threw at it, 3b models all the way to one trillion params. watch this happen on my own machine. i pointed hermes agent at a local http endpoint, gemma 4 12b on my 3090 llama.cpp server, and it auto-detected the model and started working immediately. no config wrestling, no broken tool calls, no babysitting the output format, i typed in a url and it just went. the whole clip is exactly that, start to finish, no errors, no retries, butter smooth. and the tool calling, the one thing that quietly breaks most local setups, works here like it's nothing. it's not the model that's flaky, it's the harness around it. hermes agent is the first agent i've run that actually gets that right. one url, one local model on one card, and it runs like a wild horse.

Sudo su

27,339 次观看 • 3 个月前

🚨 I'M SORRY BUT THIS IS GOING TO BE HARD BTC is down ~49% from its ATH. And most people think the worst is over. It isn't. Not yet. Two phases of this bear are done. The third, that builds every bottom - just starting And it looks nothing like what you're expecting.. Here's the roadmap, with the levels that actually matter: $60 - 61K is the line in the sand. It's held every test so far. Lose it, and the next stop isn't close. Mid-$50s. Then the real magnet: the $50–55K zone. That's where the 200-week average and the network's realized price both sit - the exact levels that marked every prior cycle bottom. That's not a random number. That's the floor cycles are built on. And the timing lines up. The four-year cycle, on-chain data, and multiple desks all point to the same window: October–December 2026. That's the peak-panic phase. When everyone swears crypto's dead. That's the bottom forming - again - right on schedule 😉 Trust me, you will not miss this market if you hit that follow button right now. Notifs on

🚨 I'M SORRY BUT THIS IS GOING TO BE HARD BTC is down ~49% from its ATH. And most people think the worst is over. It isn't. Not yet. Two phases of this bear are done. The third, that builds every bottom - just starting And it looks nothing like what you're expecting.. Here's the roadmap, with the levels that actually matter: $60 - 61K is the line in the sand. It's held every test so far. Lose it, and the next stop isn't close. Mid-$50s. Then the real magnet: the $50–55K zone. That's where the 200-week average and the network's realized price both sit - the exact levels that marked every prior cycle bottom. That's not a random number. That's the floor cycles are built on. And the timing lines up. The four-year cycle, on-chain data, and multiple desks all point to the same window: October–December 2026. That's the peak-panic phase. When everyone swears crypto's dead. That's the bottom forming - again - right on schedule 😉 Trust me, you will not miss this market if you hit that follow button right now. Notifs on

Frogify

435,176 次观看 • 1 个月前

Jensen Huang just told Silicon Valley it’s fighting on the wrong floor. Every boardroom in tech is locked on the same question. Which model wins. OpenAI or xAI. GPT or Claude. Grok or Gemini. Trillions moving on that bet alone. Huang zoomed out and showed them the whole building. Huang: “AI is actually essentially a five-layer cake.” Energy at the bottom. Chips above it. Cloud above that. Models next. Applications on top. Five layers. One war. Everyone crowded onto the fourth floor. Huang: “This is where most people think AI is.” He was pointing at the model layer. Every pitch deck. Every valuation. Every founder story. All packed onto one floor. One floor below the finish line. Three above the foundation. The middle of the building. Huang: “At the bottom is energy.” Not data. Not parameters. Not talent. Power. You cannot out-code the grid. You cannot train a frontier model with a press release. The smartest model on Earth still needs a dumb turbine spinning somewhere. The smartest engineers alive are building on top of someone else’s silicon, inside someone else’s cloud, powered by someone else’s electricity. They own nothing beneath them. Huang: “This layer on top ultimately is where economic benefit will happen.” Healthcare. Finance. Manufacturing. The only floors where AI actually meets money. Every dollar of real value lives at the top. Every physical constraint that decides who gets to play lives at the bottom. The model sits in between. Squeezed from above and below and owning neither end. Silicon Valley is burning hundreds of billions to build plumbing for somebody else’s economy. The basement decides if it runs. The penthouse decides if it pays. The companies building models think they are building the future. Huang just told them they are the middle layer in someone else’s cake.

Jensen Huang just told Silicon Valley it’s fighting on the wrong floor. Every boardroom in tech is locked on the same question. Which model wins. OpenAI or xAI. GPT or Claude. Grok or Gemini. Trillions moving on that bet alone. Huang zoomed out and showed them the whole building. Huang: “AI is actually essentially a five-layer cake.” Energy at the bottom. Chips above it. Cloud above that. Models next. Applications on top. Five layers. One war. Everyone crowded onto the fourth floor. Huang: “This is where most people think AI is.” He was pointing at the model layer. Every pitch deck. Every valuation. Every founder story. All packed onto one floor. One floor below the finish line. Three above the foundation. The middle of the building. Huang: “At the bottom is energy.” Not data. Not parameters. Not talent. Power. You cannot out-code the grid. You cannot train a frontier model with a press release. The smartest model on Earth still needs a dumb turbine spinning somewhere. The smartest engineers alive are building on top of someone else’s silicon, inside someone else’s cloud, powered by someone else’s electricity. They own nothing beneath them. Huang: “This layer on top ultimately is where economic benefit will happen.” Healthcare. Finance. Manufacturing. The only floors where AI actually meets money. Every dollar of real value lives at the top. Every physical constraint that decides who gets to play lives at the bottom. The model sits in between. Squeezed from above and below and owning neither end. Silicon Valley is burning hundreds of billions to build plumbing for somebody else’s economy. The basement decides if it runs. The penthouse decides if it pays. The companies building models think they are building the future. Huang just told them they are the middle layer in someone else’s cake.

Dustin

536,583 次观看 • 5 个月前

Jensen Huang was asked whether AI's future is open source or proprietary the question the whole industry keeps fighting over. He didn't pick a side. He said something that quietly ends the fight. "models is a technology, not a product. Model is a technology, not a service" "For the vast majority of consumers, the horizontal layer, the general intelligence, I would really, really love not to go fine-tune my own" "all these industries, their domain expertise, their specialization has to be channeled, has to be captured in a way that they can control, and that it can only come from open models" "the open model industry we're contributing tremendously to it is near the frontier" Everyone else is arguing about which lab wins OpenAI or Anthropic, who has the smartest frontier model this quarter. Jensen is standing on stage saying that's the sideshow. The proprietary model is the convenience layer you rent when you don't care. The open model is the thing you own when the work is yours. Two different products for two different needs not two horses in one race. open models are the second-biggest in the room. Ahead of Anthropic. Behind only OpenAI.

Jensen Huang was asked whether AI's future is open source or proprietary the question the whole industry keeps fighting over. He didn't pick a side. He said something that quietly ends the fight. "models is a technology, not a product. Model is a technology, not a service" "For the vast majority of consumers, the horizontal layer, the general intelligence, I would really, really love not to go fine-tune my own" "all these industries, their domain expertise, their specialization has to be channeled, has to be captured in a way that they can control, and that it can only come from open models" "the open model industry we're contributing tremendously to it is near the frontier" Everyone else is arguing about which lab wins OpenAI or Anthropic, who has the smartest frontier model this quarter. Jensen is standing on stage saying that's the sideshow. The proprietary model is the convenience layer you rent when you don't care. The open model is the thing you own when the work is yours. Two different products for two different needs not two horses in one race. open models are the second-biggest in the room. Ahead of Anthropic. Behind only OpenAI.

Vikram M

12,078 次观看 • 2 个月前

Orchestrators vs. Graphs, clearly explained! orchestrators are great, and everyone builds one first. here is the ceiling: an orchestrator sits above the work and routes every message. five agents report to it. it reads all five. it decides what each one does next, and reads all five replies. that is ten trips through one context, and by the fifth agent that context has read four reports, five instructions and its own reasoning about all of them. Graph engineering fixes this by removing the seat: not a better router, but no router at all. you need both, and here is the sentence that resolves the whole confusion: an orchestrator sits above the work and holds all of it. a graph is the shape of the work, and holds none of it. ↳ above the work: one context that has to see everything before anything ships ↳ inside the work: a splitter that hands out and lets go, and a merge that reads nothing Prompts → Context → Harness → Loops → Graphs the coordination did not disappear. it moved into the edges, where it costs nothing and cannot get tired. the trick is noticing what you actually built. if one node has to see every result before the run can finish, you did not remove the bottleneck. you hired it, gave it the longest context in the system, and made it the thing you were counting on to stay sharp. one thing to know before you scale it. an orchestrator degrades in the one way nothing catches. ↳ it does not crash, time out or return an error. it stays up and keeps routing ↳ it just starts routing worse, somewhere around the fifth report, and every downstream agent does exactly what it was told that last one catches careful people. you can have perfect isolation on every worker and still have one window quietly drifting at the top, and the traces will all look clean because each worker did its job. and the one that eats whole nights: the merge is where this shows up first. ranking five findings is not judgment, it is a sort. if a model is doing it, you are paying a model to read five reports so it can put them in an order that three lines of code would have got right, and now that model has read everything too. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this and read it below ↓

Orchestrators vs. Graphs, clearly explained! orchestrators are great, and everyone builds one first. here is the ceiling: an orchestrator sits above the work and routes every message. five agents report to it. it reads all five. it decides what each one does next, and reads all five replies. that is ten trips through one context, and by the fifth agent that context has read four reports, five instructions and its own reasoning about all of them. Graph engineering fixes this by removing the seat: not a better router, but no router at all. you need both, and here is the sentence that resolves the whole confusion: an orchestrator sits above the work and holds all of it. a graph is the shape of the work, and holds none of it. ↳ above the work: one context that has to see everything before anything ships ↳ inside the work: a splitter that hands out and lets go, and a merge that reads nothing Prompts → Context → Harness → Loops → Graphs the coordination did not disappear. it moved into the edges, where it costs nothing and cannot get tired. the trick is noticing what you actually built. if one node has to see every result before the run can finish, you did not remove the bottleneck. you hired it, gave it the longest context in the system, and made it the thing you were counting on to stay sharp. one thing to know before you scale it. an orchestrator degrades in the one way nothing catches. ↳ it does not crash, time out or return an error. it stays up and keeps routing ↳ it just starts routing worse, somewhere around the fifth report, and every downstream agent does exactly what it was told that last one catches careful people. you can have perfect isolation on every worker and still have one window quietly drifting at the top, and the traces will all look clean because each worker did its job. and the one that eats whole nights: the merge is where this shows up first. ranking five findings is not judgment, it is a sort. if a model is doing it, you are paying a model to read five reports so it can put them in an order that three lines of code would have got right, and now that model has read everything too. below i have quoted my full guide on graph engineering. it covers the three topologies, the verifier patterns, and where the gate should actually open. save this and read it below ↓

Hanako

43,781 次观看 • 10 天前

9 out of 10 multi-agent projects never leave demo mode Not because the model is bad. Because the structure is missing. Most people who try to build [ a team of AI agents ] end up with one agent talking to itself in five tabs. The agents don't share context → Don't divide work → Don't know what the others are doing. > stage 1: if your agent doesn't have a real loop, observe, act, iterate, you have a long prompt, not an agent > stage 2: subagents need isolated context, the orchestrator never reads their raw transcript, only the summary > stage 3: the orchestrator plans and delegates, the moment it executes, it drowns in details that belong inside subagents > stage 4: without a shared task list it's not a team, it's five agents duplicating each other's work in parallel > stage 5: a permissions file is what lets you sleep, the model cannot bypass it because the rule lives outside the model a team of AI agents is not more model, it is more structure the most ignored stage in every demo that died: - durability... when a 50-step task crashes at step 47 and starts from zero, that's not a model failure. that's a missing write-to-disk call

9 out of 10 multi-agent projects never leave demo mode Not because the model is bad. Because the structure is missing. Most people who try to build [ a team of AI agents ] end up with one agent talking to itself in five tabs. The agents don't share context → Don't divide work → Don't know what the others are doing. > stage 1: if your agent doesn't have a real loop, observe, act, iterate, you have a long prompt, not an agent > stage 2: subagents need isolated context, the orchestrator never reads their raw transcript, only the summary > stage 3: the orchestrator plans and delegates, the moment it executes, it drowns in details that belong inside subagents > stage 4: without a shared task list it's not a team, it's five agents duplicating each other's work in parallel > stage 5: a permissions file is what lets you sleep, the model cannot bypass it because the rule lives outside the model a team of AI agents is not more model, it is more structure the most ignored stage in every demo that died: - durability... when a 50-step task crashes at step 47 and starts from zero, that's not a model failure. that's a missing write-to-disk call

Shadow Nick

14,714 次观看 • 3 个月前

I gave JEV one prompt: "turn $24.80 into 1000x or I'm pulling the plug" it didn't ask questions. didn't negotiate. didn't say "that's unrealistic" it just started $24.80 to $31,847.52 overnight I woke up, opened Meridian Desk, and the number was already there. green, pulsing, real here's the part nobody talks about: I didn't mass-trade, didn't watch charts at 3am, didn't mass-subscribe to signal groups I built a terminal called Meridian Desk and let Jev run it while I slept Jev is the brain. it doesn't trade, it thinks it reads the entire market state every few seconds and outputs one structured decision: buy, copy, skip, or pass, each one tagged with a confidence score below Jev sit four agents that do the actual work: SCOUT finds fresh mints before anyone's even talking about them LEDGER tracks smart wallets and mirrors the ones with 90%+ win rates PULSE reads momentum and social sentiment in real time FLUX maps liquidity routes so every fill gets the best execution then there's WARDEN, the fifth agent, the last gate every decision Jev makes has to pass through WARDEN before a single dollar moves. it checks for honeypots, arms stop-losses, sizes the position, and vetoes anything that smells off if WARDEN says no, nothing happens. period at hour two the balance dropped to $7 Jev went quiet. stopped entering. WARDEN was blocking everything. confidence too low, liquidity too thin, signals disagreed it waited then around hour four something shifted. SCOUT flagged a mint nobody had seen yet, LEDGER confirmed three smart wallets were already in, PULSE showed social mentions climbing, FLUX found a clean route with under 2% slippage Jev lit up: BUY, conf 0.91 WARDEN cleared it. position filled in 1.6 milliseconds that was the first real trade. the balance went from $7 to $40 in minutes after that it just kept compounding. trade after trade, all night, no human input 96.4% copy accuracy. five agents running in sync. one decision engine calling the shots I got early access to JEV. that's it. manual research trying to do what Jev does in seconds? $5,000 a month minimum, and you'd still miss the window this isn't a bot that sprays trades and hopes. this is an architecture. intelligence layer on top, execution layer below, risk gate in between Jev analyzes. agents execute. WARDEN protects I haven't touched it since I hit start bookmark this. follow me if you want to see the terminal live

I gave JEV one prompt: "turn $24.80 into 1000x or I'm pulling the plug" it didn't ask questions. didn't negotiate. didn't say "that's unrealistic" it just started $24.80 to $31,847.52 overnight I woke up, opened Meridian Desk, and the number was already there. green, pulsing, real here's the part nobody talks about: I didn't mass-trade, didn't watch charts at 3am, didn't mass-subscribe to signal groups I built a terminal called Meridian Desk and let Jev run it while I slept Jev is the brain. it doesn't trade, it thinks it reads the entire market state every few seconds and outputs one structured decision: buy, copy, skip, or pass, each one tagged with a confidence score below Jev sit four agents that do the actual work: SCOUT finds fresh mints before anyone's even talking about them LEDGER tracks smart wallets and mirrors the ones with 90%+ win rates PULSE reads momentum and social sentiment in real time FLUX maps liquidity routes so every fill gets the best execution then there's WARDEN, the fifth agent, the last gate every decision Jev makes has to pass through WARDEN before a single dollar moves. it checks for honeypots, arms stop-losses, sizes the position, and vetoes anything that smells off if WARDEN says no, nothing happens. period at hour two the balance dropped to $7 Jev went quiet. stopped entering. WARDEN was blocking everything. confidence too low, liquidity too thin, signals disagreed it waited then around hour four something shifted. SCOUT flagged a mint nobody had seen yet, LEDGER confirmed three smart wallets were already in, PULSE showed social mentions climbing, FLUX found a clean route with under 2% slippage Jev lit up: BUY, conf 0.91 WARDEN cleared it. position filled in 1.6 milliseconds that was the first real trade. the balance went from $7 to $40 in minutes after that it just kept compounding. trade after trade, all night, no human input 96.4% copy accuracy. five agents running in sync. one decision engine calling the shots I got early access to JEV. that's it. manual research trying to do what Jev does in seconds? $5,000 a month minimum, and you'd still miss the window this isn't a bot that sprays trades and hopes. this is an architecture. intelligence layer on top, execution layer below, risk gate in between Jev analyzes. agents execute. WARDEN protects I haven't touched it since I hit start bookmark this. follow me if you want to see the terminal live

cristal💎

225,879 次观看 • 2 天前

ANTHROPIC'S CEO SAYS AI IS MAYBE 6 TO 12 MONTHS FROM DOING EVERYTHING ENGINEERS DO. Two types of people. Type 1 learns to direct the systems. Type 2 waits and gets directed out. The timeline is uncertain. The direction is not. Control the system or become what it replaces. Bookmark this. Follow CyrilXBT

ANTHROPIC'S CEO SAYS AI IS MAYBE 6 TO 12 MONTHS FROM DOING EVERYTHING ENGINEERS DO. Two types of people. Type 1 learns to direct the systems. Type 2 waits and gets directed out. The timeline is uncertain. The direction is not. Control the system or become what it replaces. Bookmark this. Follow CyrilXBT

CyrilXBT

14,763 次观看 • 2 个月前

Jev is cool. So is it's OSS companion, Laya. The Latest Cool Thing In AI™ tends to get a lot of hype, sometimes without everyone even understanding it. So... what is this thing? Jev is an AI model that consumes input and produces output VERY differently than chat, claude, grok. The input is two things: 1) Text state to assess. Email, html, code, whatever. 2) A set of questions which will be asked about the attached state. The canonical example from TypeSafe's docs is to identify the urgency of a support ticket. We pass the model the customer text + a single noul question "is this urgent?". Jev returns a full set of JSON. This JSON is not generated with token-by-token autoregression. Jev is not trained to produce sequences of text tokens, rather to answer questions, and guarantees well-formed responses. In the example below, we see it produces a 0.99 probability (on a 0-1.0 scale) that the answer is "yes." Jev supports exactly three types of questions (seconds example in video): a) Noul: 0–1 probability that the answer to a yes/no question is "yes." b) Choice: Ask question with pre-defined set of answers. Jev chooses the best and assigns probabilities to each. c) Score: Ask question with pre-defined scale of answers. Jev produces a position on the scale. Jev computes answers for all questions in parallel, making responses super fast even for many questions in a single request. This might seem like a narrow set of capabilities, but in the right contexts leads to incredible potential. It also makes for a useful API / primitive for programming, since the outputs are... *ahem*... type-safe and predictable in structure. Jev is not going to replace LLMs for writing your code, auto-generating your docs, or being at the core of an agent harness. But Jev IS incredibly cool, and will be used to build a lot of amazing tech. Hope this helps.

Jev is cool. So is it's OSS companion, Laya. The Latest Cool Thing In AI™ tends to get a lot of hype, sometimes without everyone even understanding it. So... what is this thing? Jev is an AI model that consumes input and produces output VERY differently than chat, claude, grok. The input is two things: 1) Text state to assess. Email, html, code, whatever. 2) A set of questions which will be asked about the attached state. The canonical example from TypeSafe's docs is to identify the urgency of a support ticket. We pass the model the customer text + a single noul question "is this urgent?". Jev returns a full set of JSON. This JSON is not generated with token-by-token autoregression. Jev is not trained to produce sequences of text tokens, rather to answer questions, and guarantees well-formed responses. In the example below, we see it produces a 0.99 probability (on a 0-1.0 scale) that the answer is "yes." Jev supports exactly three types of questions (seconds example in video): a) Noul: 0–1 probability that the answer to a yes/no question is "yes." b) Choice: Ask question with pre-defined set of answers. Jev chooses the best and assigns probabilities to each. c) Score: Ask question with pre-defined scale of answers. Jev produces a position on the scale. Jev computes answers for all questions in parallel, making responses super fast even for many questions in a single request. This might seem like a narrow set of capabilities, but in the right contexts leads to incredible potential. It also makes for a useful API / primitive for programming, since the outputs are... ahem... type-safe and predictable in structure. Jev is not going to replace LLMs for writing your code, auto-generating your docs, or being at the core of an agent harness. But Jev IS incredibly cool, and will be used to build a lot of amazing tech. Hope this helps.

Ben Dicken

40,810 次观看 • 5 天前