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。

👶🤖 Introducing `gpt-engineer` ▸ One prompt generates a codebase ▸ Asks clarifying questions ▸ Generates technical spec ▸ Writes all necessary code ▸ Easy to add your own reasoning steps, modify, and experiment ▸ open source: ▸ Lets you finish a coding project in minutes. swyx scooped me on... show more

Anton Osika

85,014 subscribers

665,005 次观看 • 3 年前 •via X (Twitter)

科学技术教育艺术

Anya Rossi• Live Now

Private livecam show

10 条评论

Alex Mizrahi 的头像

Alex Mizrahi3 年前

Generating a whole code base at once is a nice trick but it works only for tiny code bases. Basically as soon as things don't fit into the context the quality will drop sharply and the tool won't be useful. To make a real GPT engineer you need to make it incremental. It should work only on a small task at a time. For example it might work in a loop: 1. Select parts of the code base which are relevant for a particular task. 2. Generate new code taking the context into consideration. 3. Update the summary of the code base. Of course, with a larger context size it might become feasible to update the whole code base at once. But I doubt that quality will be the same as with a smaller context. And you hit a hard limit once you exceed the context size.

Sid⚡️ 的头像

Sid⚡️3 年前

Bonus feature: Over promise and under deliver like a true consulting engineer

spink 的头像

spink3 年前

This looks cool! I’m working on something pretty similar

SQUIP 的头像

SQUIP3 年前

this is amazing 🔥🔥. . any tips on how to make this work with chatgpt-3.5 turbo for the plebs on the waitlist for gpt4 api access? I'm getting an error bc of the context length token limits of the 3.5 model. I want to experiment with this soo bad.

Daniel Deveney 的头像

Daniel Deveney3 年前

But did the code work? And does it have the ability to clarify that it works, and debug it if it doesn’t?

Mihai ᛋ 的头像

Mihai ᛋ3 年前

bro why didn't you hit run

Jacob Valdez 的头像

Jacob Valdez3 年前

Thank you thank you thank you But at the same time 😭😭😭😭😭😭😭. There goes my profession

Robin Ranjit Singh Chauhan @ Neurips 的头像

Robin Ranjit Singh Chauhan @ Neurips3 年前

So did the snake game work? Demo ended for it ran. Anyways cool project!

dj spry 的头像

dj spry3 年前

@SteveMoraco @Memdotai mem it

Mem 的头像

Mem3 年前

@antonosika @SteveMoraco Saved! Here's the compiled thread: 🪄 AI-generated summary: "gpt-engineer is an open source tool that allows users to generate a codebase, ask clarifying questions, create a technical spec, and write all necessary code with ease. It also...

相关视频

ONE PROMPT AND ONE BUTTON NOW BUILDS A FULL SYSTEM, DATABASE, BACKEND, AND FRONTEND INCLUDED Most people still picture AI as something you ask questions to, then copy the answer into your own code by hand. A builder wired Claude into Obsidian through MCP instead, so a single written prompt becomes the actual spec for the system. Claude breaks that prompt into database schema, UI components, and API endpoints on its own, then writes, runs, and fixes the code until it actually works. MCP is what makes this possible, it lets the agent see the filesystem, installed packages, and running services before it starts building anything. The output isn't a code snippet sitting in a chat window, it's a live local app running on its own server, ready to open in a browser. What used to need a backend developer, a DevOps engineer, and two weeks now finishes in under fifteen minutes. See the full architecture and prompt structure below👇

ONE PROMPT AND ONE BUTTON NOW BUILDS A FULL SYSTEM, DATABASE, BACKEND, AND FRONTEND INCLUDED Most people still picture AI as something you ask questions to, then copy the answer into your own code by hand. A builder wired Claude into Obsidian through MCP instead, so a single written prompt becomes the actual spec for the system. Claude breaks that prompt into database schema, UI components, and API endpoints on its own, then writes, runs, and fixes the code until it actually works. MCP is what makes this possible, it lets the agent see the filesystem, installed packages, and running services before it starts building anything. The output isn't a code snippet sitting in a chat window, it's a live local app running on its own server, ready to open in a browser. What used to need a backend developer, a DevOps engineer, and two weeks now finishes in under fifteen minutes. See the full architecture and prompt structure below👇

wast3

20,348 次观看 • 3 个月前

Today we're launching Nova - an AI software engineer by Supercode. an oss alternative for Devin by Cognition give it a task. It reads your codebase, plans the change, writes the code, runs your tests, reviews its own work, and reports back like a teammate would. a harness, an AI code review, and voice agent all in one system. completely open source. we're launching it on early personal invites. comment Nova or DM me for early access.

Today we're launching Nova - an AI software engineer by Supercode. an oss alternative for Devin by Cognition give it a task. It reads your codebase, plans the change, writes the code, runs your tests, reviews its own work, and reports back like a teammate would. a harness, an AI code review, and voice agent all in one system. completely open source. we're launching it on early personal invites. comment Nova or DM me for early access.

Yash

16,598 次观看 • 1 个月前

PROMPTS ARE DEAD. LOOPS JUST REPLACED YOUR $200K ENGINEER. Claire Vo showed how to turn one prompt into a self-running loop. It reviews your aging pull requests every morning and hunts for new skills every week. And it spins up its own subagents to check its own work. Runs on Claude Code + Codex. One afternoon. One prompt. The numbers are insane: - Senior doing code review: $200K/year - Automation engineer: $150K/year - This setup: about $200/month - Time to build: one afternoon - Skill required: describe the job like you are onboarding a hire The opportunity is even wilder: Teams pay people to babysit pull requests and chase stale reviews. The loop does it itself at 10:15 a.m., alerts the team, and validates itself with subagents. An agency would charge thousands to wire this up. One person + Claude Code + Codex + one prompt = an engineering team that reviews code and finds new skills while you sleep. Save and watch the clip.

PROMPTS ARE DEAD. LOOPS JUST REPLACED YOUR $200K ENGINEER. Claire Vo showed how to turn one prompt into a self-running loop. It reviews your aging pull requests every morning and hunts for new skills every week. And it spins up its own subagents to check its own work. Runs on Claude Code + Codex. One afternoon. One prompt. The numbers are insane: - Senior doing code review: $200K/year - Automation engineer: $150K/year - This setup: about $200/month - Time to build: one afternoon - Skill required: describe the job like you are onboarding a hire The opportunity is even wilder: Teams pay people to babysit pull requests and chase stale reviews. The loop does it itself at 10:15 a.m., alerts the team, and validates itself with subagents. An agency would charge thousands to wire this up. One person + Claude Code + Codex + one prompt = an engineering team that reviews code and finds new skills while you sleep. Save and watch the clip.

shmidt

51,155 次观看 • 3 个月前

The idea isn't "make original content." It's find what's already working, hand it to GPT-6 Astra to reverse-engineer, then let Postiz handle the boring part. This setup is pulling +26.4M views a week! > The research: TikTok's own like-filters show what's repeatably viral, not what got lucky once > The AI's real job: GPT-6 Astra doesn't generate ideas from nothing, it extracts the pattern from proven slideshows first > The output: one winning structure becomes 20 new concepts across totally different stories > The production line: GPT Image 2.5 batch-generates a week's worth of slideshows in one sitting > The distribution: a ChatGPT agent plans the week, Postiz schedules and publishes it > The discipline: winners get variations, anything under 3k views gets cut immediately Watch this, then read the full breakdown below.

The idea isn't "make original content." It's find what's already working, hand it to GPT-6 Astra to reverse-engineer, then let Postiz handle the boring part. This setup is pulling +26.4M views a week! > The research: TikTok's own like-filters show what's repeatably viral, not what got lucky once > The AI's real job: GPT-6 Astra doesn't generate ideas from nothing, it extracts the pattern from proven slideshows first > The output: one winning structure becomes 20 new concepts across totally different stories > The production line: GPT Image 2.5 batch-generates a week's worth of slideshows in one sitting > The distribution: a ChatGPT agent plans the week, Postiz schedules and publishes it > The discipline: winners get variations, anything under 3k views gets cut immediately Watch this, then read the full breakdown below.

NeilXbt

13,798 次观看 • 1 个月前

ONE PROMPT IN CHATGPT IMAGE 2 JUST REPLACED AN ENTIRE $5,000-$50,000 BRAND DESIGN TEAM Full brand identity. Colors, typography, logo style, aesthetic. All generated in one single prompt. Here's the exact workflow: Step 1: Paste one prompt asking for a complete brand identity. GPT Image 2 generates everything – color palette, fonts, logo direction, overall visual style. Step 2: Use that output as context for every asset you generate after. Campaign visuals, shop graphics, product mockups, social posts – all stay consistent and on-brand automatically. The result: A full visual identity that looks like a real design team built it. Clean text rendering. Consistent style across every single asset. Design agencies charge $5,000-$50,000 for brand identity work. GPT Image 2 does it in one prompt. The full prompt that makes this work is in the video below. Bookmark this post, you'll want to come back to it.

SCOTTY BEAM

12,023 次观看 • 4 个月前

ANTHROPIC LEAKED A SETUP WHERE ONE FILE STARTS 7 LOOPS THAT TURN A PROMPT INTO A WORKING BUSINESS AI engineer drops in 1 file - the other 95% of the work closes automatically without them. spec → tasks → code → tests → docs → release → monitoring → back into the spec you don't write tickets anymore - you write the thing tickets get derived from, and Opus 5 cuts the spec into 6 tasks by itself. a task that fails its test comes back to itself, not to you - and runs a second lap without a single notification. 6 lanes of code run in parallel and none of them wait on you - the critic filters the diff before you ever open it. every new production bug becomes a permanent line in the eval set - the system never steps on it twice. monitoring isn't the end of the line, it's the start of the next spec - those back edges are what make it a graph instead of a conveyor. and one place is left where you say yes - release, 5 minutes instead of 5 hours. save this and paste it into Claude Code - and you get 7 loops that run without you ↓

ANTHROPIC LEAKED A SETUP WHERE ONE FILE STARTS 7 LOOPS THAT TURN A PROMPT INTO A WORKING BUSINESS AI engineer drops in 1 file - the other 95% of the work closes automatically without them. spec → tasks → code → tests → docs → release → monitoring → back into the spec you don't write tickets anymore - you write the thing tickets get derived from, and Opus 5 cuts the spec into 6 tasks by itself. a task that fails its test comes back to itself, not to you - and runs a second lap without a single notification. 6 lanes of code run in parallel and none of them wait on you - the critic filters the diff before you ever open it. every new production bug becomes a permanent line in the eval set - the system never steps on it twice. monitoring isn't the end of the line, it's the start of the next spec - those back edges are what make it a graph instead of a conveyor. and one place is left where you say yes - release, 5 minutes instead of 5 hours. save this and paste it into Claude Code - and you get 7 loops that run without you ↓

Sprytix

66,333 次观看 • 2 个月前

Nothing beats Augment Code answering questions about large codebases. It's a free extension (works in VSCode, JetBrains, and NeoVim), and has a community edition you can run for free forever. Try this experiment: 1. Download any large open-source project from GitHub 2. Augment will index the code 3. Start asking questions about the codebase 4. Do the same using any other coding agent 5. Compare the speed and the depth of the answers you get If you have a small codebase, the differences with other coding agents won't be obvious, but if you are working on a large project, things will click immediately. Suggestion: download the Keras repository (~180,000 lines of code). I know the library well, and the answers I got from Augment were pretty impressive. By the way, Augment is a fully-fledged AI coding assistant. You can use its agent in every imaginable way: • To implement brand new functionality • To make quick edits to your code • To refactor existing functionality • To generate unit tests • To generate documentation • To handle GitHub—pull requests, commits, branching, etc My favorite feature is their "Next Edit" functionality: 1. You edit the code 2. Augment analyzes the change (across the entire codebase!) 3. Augment determines the ripple effects of that change 4. Augment suggests everything you need to update Here is a link to the tool: Thanks to the team at Augment Code for partnering with me on this post.

Santiago

34,977 次观看 • 1 年前

Hi AI Engineer @ Paris 🇫🇷, we turned one of your talks into something you can actually post. We gave Pexo one of your session replays. It watched the whole talk and chose the format on its own. An hour of dense, technical ideas, drawn out step by step on a whiteboard. The best parts of the talk, now easy to share. Which talk should Pexo take on next? #MadeWithPexo

Hi AI Engineer @ Paris 🇫🇷, we turned one of your talks into something you can actually post. We gave Pexo one of your session replays. It watched the whole talk and chose the format on its own. An hour of dense, technical ideas, drawn out step by step on a whiteboard. The best parts of the talk, now easy to share. Which talk should Pexo take on next? #MadeWithPexo

Pexo

24,161 次观看 • 27 天前

This workflow combining Loom’s AI features + a custom ChatGPT GPT is saving me hours. Instead of creating onboarding Docs for new team members, I film a Loom → generate SOP → train a GPT to answer questions Game changer for businesses to delegate faster. Here's how to do it: First, I record a video of whatever task I want to delegate to the new team member with Loom. The more in-depth, the better, but I just used a 7-minute video. Then, I use Loom's new AI 'Write a document' feature. Upgrading to Loom AI from the standard Loom plan cost me $2. Loom AI can generate an entire SOP, PR description, step-by-step guide, QA, and more from a simple Loom video in <5 seconds. In the past, I’ve spent 2 hours+ hand-writing each one of these docs to onboard new team members, so Loom AI is already a massive timesaver. But it gets even better! Next, we can take that data from the SOP document, and we use it as 'Knowledge' to train a Custom GPT that can answer the new team members' questions. The more SOP docs/Knowledge you feed the GPT, the better. But one is fine if that's all you have because the GPT will pull any unknown answers from the web or its training data. Here are the prompt Instructions you want to put into the Custom GPT (copy and paste this): You are an expert CEO, specialized in onboarding and training new team members. Using the Knowledge provided, you will help new team members with any questions or stipulations they may have about their new role. Stick as true to the data provided as possible, but if= they ask any questions that the Knowledge base does not have a specific answer for, you are permitted to use your pre-trained data and/or web browsing capabilities. That's it! It can't replace you entirely, but it'll save you 90% of the time you would've wasted on writing an SOP doc and answering questions. Simple AI workflows here and there really add up. There's also a workflow to help with the job screening process, but I'll save that for another day :^)

This workflow combining Loom’s AI features + a custom ChatGPT GPT is saving me hours. Instead of creating onboarding Docs for new team members, I film a Loom → generate SOP → train a GPT to answer questions Game changer for businesses to delegate faster. Here's how to do it: First, I record a video of whatever task I want to delegate to the new team member with Loom. The more in-depth, the better, but I just used a 7-minute video. Then, I use Loom's new AI 'Write a document' feature. Upgrading to Loom AI from the standard Loom plan cost me $2. Loom AI can generate an entire SOP, PR description, step-by-step guide, QA, and more from a simple Loom video in <5 seconds. In the past, I’ve spent 2 hours+ hand-writing each one of these docs to onboard new team members, so Loom AI is already a massive timesaver. But it gets even better! Next, we can take that data from the SOP document, and we use it as 'Knowledge' to train a Custom GPT that can answer the new team members' questions. The more SOP docs/Knowledge you feed the GPT, the better. But one is fine if that's all you have because the GPT will pull any unknown answers from the web or its training data. Here are the prompt Instructions you want to put into the Custom GPT (copy and paste this): You are an expert CEO, specialized in onboarding and training new team members. Using the Knowledge provided, you will help new team members with any questions or stipulations they may have about their new role. Stick as true to the data provided as possible, but if= they ask any questions that the Knowledge base does not have a specific answer for, you are permitted to use your pre-trained data and/or web browsing capabilities. That's it! It can't replace you entirely, but it'll save you 90% of the time you would've wasted on writing an SOP doc and answering questions. Simple AI workflows here and there really add up. There's also a workflow to help with the job screening process, but I'll save that for another day :^)

Rowan Cheung

129,424 次观看 • 2 年前

Claude Code is unusable. So I built my own. Introducing BridgeCode. Launching next Wednesday. Everything you love about Claude Code. No rate limits. No 529 errors. No 500 errors. No nerfing. Connect to any provider. Claude Opus 4.6. GPT 5.4. Gemini 3.1 Pro. GLM 5.1. All of them. Switch models in seconds. And one more thing. Swarm mode. One prompt. Ten agents. Working together to ship your code. Claude Code rate limited me into building a competitor. Thank you, Anthropic. Launching next Wednesday.

Claude Code is unusable. So I built my own. Introducing BridgeCode. Launching next Wednesday. Everything you love about Claude Code. No rate limits. No 529 errors. No 500 errors. No nerfing. Connect to any provider. Claude Opus 4.6. GPT 5.4. Gemini 3.1 Pro. GLM 5.1. All of them. Switch models in seconds. And one more thing. Swarm mode. One prompt. Ten agents. Working together to ship your code. Claude Code rate limited me into building a competitor. Thank you, Anthropic. Launching next Wednesday.

BridgeMind

47,273 次观看 • 6 个月前

I just asked Claude Fable 5 to show me its muscles and build the best game it could with PRD and Goal in 1 shot Done, Torched 90% of my tokens in two 5 hours sessions 🤦 ( it spend 2x the token than opus confirmed) Come admire Claude's 5 hour master piece - an ink-wash calligraphy defense game with roguelike, skills, bossfight and end game delight. Your finger IS the brush: any stroke is a blade. What I found: - Fables is a lot more like a Artist during desiging and have its own taste. it created all the game art assets and even the guqin music. - During testing, it is also showing GPT 5.5 level testing, fired up Playwright on its own, screenshot-tested every boss, caught and fixed its own bugs if GPT 5.5 is a solid engineer, Fables 5 is a Artist and a solid Engineer all in one check out the game in comments 👇

I just asked Claude Fable 5 to show me its muscles and build the best game it could with PRD and Goal in 1 shot Done, Torched 90% of my tokens in two 5 hours sessions 🤦 ( it spend 2x the token than opus confirmed) Come admire Claude's 5 hour master piece - an ink-wash calligraphy defense game with roguelike, skills, bossfight and end game delight. Your finger IS the brush: any stroke is a blade. What I found: - Fables is a lot more like a Artist during desiging and have its own taste. it created all the game art assets and even the guqin music. - During testing, it is also showing GPT 5.5 level testing, fired up Playwright on its own, screenshot-tested every boss, caught and fixed its own bugs if GPT 5.5 is a solid engineer, Fables 5 is a Artist and a solid Engineer all in one check out the game in comments 👇

Nil Ni 👶🦊🗝🎮🐷💎🎆

413,328 次观看 • 4 个月前

ChatGPT "Project Sunshine" is ChatGPT with Memory & New GPT Store Layout ChatGPT's "Project Sunshine," which was initially discovered in August 2023 as a version of ChatGPT with special capabilities, is actually ChatGPT with Memory. The current version of ChatGPT features a new "Personalization" section in its settings: ``` Improve responses with your chats Your GPT will continually improve as you chat, picking up on details and preferences to tailor its responses to you. Learn more To modify what your GPT knows or teach it something new, just tell it: “What do you know about me?” “Where did we leave off on my last project?” “Remember that I like concise responses.” ``` This section includes options to turn the feature on or off and to "Reset Memory." The current status of personalization is also displayed as "Memory Off" on the dashboard, indicating that "ChatGPT won't remember anything you talk about in this conversation." Additionally, the new ChatGPT with Memory feature is introduced via a new popup: ``` Your GPT can now learn from your chats Keep the conversation going Your GPT will carry what it learns between chats, allowing it to provide more relevant responses. Improves over time As you chat your GPT will become more helpful, remembering details and preferences. Manage what it remembers Your GPT has been designed to follow your instructions in chats. You can reset your GPT's memory or turn this feature off in settings. ``` The previously mentioned "Temporary Chat" is now visible in the model selector (however, it appears to be a bit buggy) and alters the style of the prompt text input, similar to the existing "Chat history & training" on/off option in Settings. "Temporary Chat" also has its own announcement popup explaining its functionality: ``` Temporary Chat Not in history Temporary Chats won't appear in your history, and your GPT won't remember anything you talk about. We may still keep a copy for up to 30 days for safety purposes. Blank slate Your GPT won't be aware of previous conversations. It will still follow your custom instructions if they're enabled. No model training Temporary Chats won't be used to improve our models. ``` Moreover, the GPT Store has received an update as it approaches its launch, featuring a prominent header, search bar, and lists of My GPTs, Recently Used, Made by OpenAI, and possibly Featured & Top GPTs soon.

ChatGPT "Project Sunshine" is ChatGPT with Memory & New GPT Store Layout ChatGPT's "Project Sunshine," which was initially discovered in August 2023 as a version of ChatGPT with special capabilities, is actually ChatGPT with Memory. The current version of ChatGPT features a new "Personalization" section in its settings: ``` Improve responses with your chats Your GPT will continually improve as you chat, picking up on details and preferences to tailor its responses to you. Learn more To modify what your GPT knows or teach it something new, just tell it: “What do you know about me?” “Where did we leave off on my last project?” “Remember that I like concise responses.” ``` This section includes options to turn the feature on or off and to "Reset Memory." The current status of personalization is also displayed as "Memory Off" on the dashboard, indicating that "ChatGPT won't remember anything you talk about in this conversation." Additionally, the new ChatGPT with Memory feature is introduced via a new popup: ``` Your GPT can now learn from your chats Keep the conversation going Your GPT will carry what it learns between chats, allowing it to provide more relevant responses. Improves over time As you chat your GPT will become more helpful, remembering details and preferences. Manage what it remembers Your GPT has been designed to follow your instructions in chats. You can reset your GPT's memory or turn this feature off in settings. ``` The previously mentioned "Temporary Chat" is now visible in the model selector (however, it appears to be a bit buggy) and alters the style of the prompt text input, similar to the existing "Chat history & training" on/off option in Settings. "Temporary Chat" also has its own announcement popup explaining its functionality: ``` Temporary Chat Not in history Temporary Chats won't appear in your history, and your GPT won't remember anything you talk about. We may still keep a copy for up to 30 days for safety purposes. Blank slate Your GPT won't be aware of previous conversations. It will still follow your custom instructions if they're enabled. No model training Temporary Chats won't be used to improve our models. ``` Moreover, the GPT Store has received an update as it approaches its launch, featuring a prominent header, search bar, and lists of My GPTs, Recently Used, Made by OpenAI, and possibly Featured & Top GPTs soon.

Tibor Blaho

194,769 次观看 • 2 年前

Today we’re excited to feature RAGApp v0.1 - which lets any user construct a multi-agent application 🎨🤖 without writing a single line of code 💫 Add any number of agents that you wish, and assign each agent a role, system prompt, and a set of tools. In this example, use a researcher, analyst, and report generation agent to write a news article. This directly generates a full chat interface where you can ask questions and get back answers with full streaming and sources. Huge shoutout to Marcus Schiesser for working on this! RAGApp: create-llama: If you want finer control, you can define your own agentic workflows through code:

Today we’re excited to feature RAGApp v0.1 - which lets any user construct a multi-agent application 🎨🤖 without writing a single line of code 💫 Add any number of agents that you wish, and assign each agent a role, system prompt, and a set of tools. In this example, use a researcher, analyst, and report generation agent to write a news article. This directly generates a full chat interface where you can ask questions and get back answers with full streaming and sources. Huge shoutout to Marcus Schiesser for working on this! RAGApp: create-llama: If you want finer control, you can define your own agentic workflows through code:

LlamaIndex 🦙

93,022 次观看 • 2 年前

What does it mean when an AI model is “open weights” and how is it different than “open source”? Welcome to the next episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! Every week we will discuss open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Episode 3 out next week; episode 1 linked below. Please leave all your questions about open-source & open-weight AI in the comments. This is a super important topic that is novel to most people, and it can be challenging to grasp - so your questions are shared with many others.

What does it mean when an AI model is “open weights” and how is it different than “open source”? Welcome to the next episode of the Open Model Series with merve, Hugging Face AI/ML engineer & open-sourceress! Every week we will discuss open-source & open-weight AI: what it is, how it works, and how you can run them on your own compute. Episode 3 out next week; episode 1 linked below. Please leave all your questions about open-source & open-weight AI in the comments. This is a super important topic that is novel to most people, and it can be challenging to grasp - so your questions are shared with many others.

Harper Carroll

32,624 次观看 • 2 个月前

This has been possible for months and barely anyone is doing it Months ago, Andrej Karpathy, a co-founder of OpenAI, open-sourced a project that gets you the best possible answers instead of trusting one model on its own. It's called LLM Council. Repo: /karpathy/llm-council The idea in two lines: your question goes to several models at once, Claude, GPT, Gemini, whatever you pick. They answer independently, then anonymously critique and rank each other (anonymous so they can't play favorites), and a chairman model merges the best of all of it into one final answer. One model just agrees with you; a council catches each other's mistakes. How to set it up: > Clone the repo > It uses uv for setup, install dependencies with uv sync > Get an OpenRouter API key and add it, this is what lets one app reach Claude, GPT, and Gemini together > Pick which models sit on your council in the config, then run it. It opens as a local web app, like ChatGPT but with a whole panel answering Bookmark this

This has been possible for months and barely anyone is doing it Months ago, Andrej Karpathy, a co-founder of OpenAI, open-sourced a project that gets you the best possible answers instead of trusting one model on its own. It's called LLM Council. Repo: /karpathy/llm-council The idea in two lines: your question goes to several models at once, Claude, GPT, Gemini, whatever you pick. They answer independently, then anonymously critique and rank each other (anonymous so they can't play favorites), and a chairman model merges the best of all of it into one final answer. One model just agrees with you; a council catches each other's mistakes. How to set it up: > Clone the repo > It uses uv for setup, install dependencies with uv sync > Get an OpenRouter API key and add it, this is what lets one app reach Claude, GPT, and Gemini together > Pick which models sit on your council in the config, then run it. It opens as a local web app, like ChatGPT but with a whole panel answering Bookmark this

Yarchi

21,488 次观看 • 3 个月前

For those of you tired of yet another AI-based IDE: Check out Traycer. It's a Visual Studio Code extension (free forever for anyone working on open-source repositories), and it does something different from everyone else. You write a high-level prompt outlining the changes you want to make to your codebase, and the extension generates a plan for you before you touch the code. (It reminds me of GitHub Copilot Workspaces.) You can review the plan, understand exactly the changes the extension will make to your code, and make any modifications. When you are ready, you execute the plan, and Traycer stages all of the changes so you have a new opportunity to review everything. Only then does Traycer apply the changes. This is ideal for large codebases where you'd rather study the changes the AI plans to make before they make it into your code. Look at the video I recorded to see how it works. Thanks to the Traycer team for collaborating with me on this post!

For those of you tired of yet another AI-based IDE: Check out Traycer. It's a Visual Studio Code extension (free forever for anyone working on open-source repositories), and it does something different from everyone else. You write a high-level prompt outlining the changes you want to make to your codebase, and the extension generates a plan for you before you touch the code. (It reminds me of GitHub Copilot Workspaces.) You can review the plan, understand exactly the changes the extension will make to your code, and make any modifications. When you are ready, you execute the plan, and Traycer stages all of the changes so you have a new opportunity to review everything. Only then does Traycer apply the changes. This is ideal for large codebases where you'd rather study the changes the AI plans to make before they make it into your code. Look at the video I recorded to see how it works. Thanks to the Traycer team for collaborating with me on this post!

Santiago

84,565 次观看 • 1 年前

OpenAI just announced API access to o1 (advanced reasoning model) yesterday. I'm delighted to announce today a new short course, Reasoning with o1, built with OpenAI, and taught by Colin Jarvis, Head of AI Solutions at OpenAI, to show you how to use this effectively! Unlike previous language models which generate output directly, o1 “thinks before it responds,” and generates many reasoning tokens before returning a more thoughtful and accurate response. It is great at complex reasoning -- including planning for agentic workflows, coding, and domain-specific reasoning in STEM fields like law. But how you should use it is quite different from other LLMs. I think o1 will be a game changer for many AI applications; and in this course, you'll learn how to use it effectively. In detail, you’ll: - Learn to recognize what tasks o1 is suited for, and when to use a smaller model, or combine o1 with a smaller model - Understand the new principles of prompting reasoning models: Be simple and direct; no explicit chain-of-thought required; use structure; show rather than tell - Implement multi-step orchestration in which o1 plans, and hands tasks over to gpt-4o-mini to execute specific steps; this illustrates a design pattern to optimize intelligence (accuracy) and cost - Use o1 for a coding task to build a new application, edit existing code, and test performance by running a coding competition between o1-mini and GPT 4o - Use o1 for image understanding and learn how it performs better with a "hierarchy of reasoning," in which it incurs the latency and cost upfront, preprocessing the image and indexing it with rich details so it can be used for Q&A later - Learn a technique called meta-prompting, in which you use o1 to improve your prompts. Using a customer support evaluation set, you'll iteratively use o1 to modify a prompt to improve performance You'll also learn about how OpenAI used reinforcement learning to produce a model that uses "test-time compute" to improve performance. I think you'll find this course enjoyable and valuable. Please sign up for it here:

OpenAI just announced API access to o1 (advanced reasoning model) yesterday. I'm delighted to announce today a new short course, Reasoning with o1, built with OpenAI, and taught by Colin Jarvis, Head of AI Solutions at OpenAI, to show you how to use this effectively! Unlike previous language models which generate output directly, o1 “thinks before it responds,” and generates many reasoning tokens before returning a more thoughtful and accurate response. It is great at complex reasoning -- including planning for agentic workflows, coding, and domain-specific reasoning in STEM fields like law. But how you should use it is quite different from other LLMs. I think o1 will be a game changer for many AI applications; and in this course, you'll learn how to use it effectively. In detail, you’ll: - Learn to recognize what tasks o1 is suited for, and when to use a smaller model, or combine o1 with a smaller model - Understand the new principles of prompting reasoning models: Be simple and direct; no explicit chain-of-thought required; use structure; show rather than tell - Implement multi-step orchestration in which o1 plans, and hands tasks over to gpt-4o-mini to execute specific steps; this illustrates a design pattern to optimize intelligence (accuracy) and cost - Use o1 for a coding task to build a new application, edit existing code, and test performance by running a coding competition between o1-mini and GPT 4o - Use o1 for image understanding and learn how it performs better with a "hierarchy of reasoning," in which it incurs the latency and cost upfront, preprocessing the image and indexing it with rich details so it can be used for Q&A later - Learn a technique called meta-prompting, in which you use o1 to improve your prompts. Using a customer support evaluation set, you'll iteratively use o1 to modify a prompt to improve performance You'll also learn about how OpenAI used reinforcement learning to produce a model that uses "test-time compute" to improve performance. I think you'll find this course enjoyable and valuable. Please sign up for it here:

Andrew Ng

358,273 次观看 • 1 年前

DELETE FIGMA. ONE MCP CONNECTOR TURNS CLAUDE CODE INTO A GOD-MODE STUDIO IN 30 MINUTES. The exact setup, step by step, no jokes: > open connectors, add a custom connector, drop in the Higgsfield MCP > open a Claude Code session and paste one prompt > it generates the first frame, the last frame, and the motion between them > then wires the result straight into an animated hero section for your site > 30 minutes end to end, nothing rendered by hand The bigger thing hiding here: MCP means the model rents hands. it doesn't know how to animate, so it hires a tool that does, mid-task, without you switching tabs. The ceiling of what one prompt ships isn't the model anymore, it's how many hands you've plugged into it. The method underneath (the goal, the rules, the bar, the loop) is in the article below, and it works with any connector you add next.

DELETE FIGMA. ONE MCP CONNECTOR TURNS CLAUDE CODE INTO A GOD-MODE STUDIO IN 30 MINUTES. The exact setup, step by step, no jokes: > open connectors, add a custom connector, drop in the Higgsfield MCP > open a Claude Code session and paste one prompt > it generates the first frame, the last frame, and the motion between them > then wires the result straight into an animated hero section for your site > 30 minutes end to end, nothing rendered by hand The bigger thing hiding here: MCP means the model rents hands. it doesn't know how to animate, so it hires a tool that does, mid-task, without you switching tabs. The ceiling of what one prompt ships isn't the model anymore, it's how many hands you've plugged into it. The method underneath (the goal, the rules, the bar, the loop) is in the article below, and it works with any connector you add next.

Miraqle

38,710 次观看 • 3 个月前