The best Hacker News stories from Show from the past day
Latest posts:
Show HN: Semantic clusters and embeddings for 500k Hacker News comments
Show HN: Semantic clusters and embeddings for 500k Hacker News comments
Show HN: Semantic clusters and embeddings for 500k Hacker News comments
Show HN: PDF to Podcast – Convert Any PDF into a Podcast Episode
Hi HN!<p>I'm stoked to share a project I've been working on called PDF to Podcast. It's a free, open-source tool that automatically converts PDF documents into engaging, informative podcast-style audio content using large language models and text-to-speech tech.<p>Inspiration:
The idea for this project came from the NotebookLM demo at Google I/O, where they showcased generating audio dialogue from uploaded PDFs and other sources. However, that audio feature hasn't been publicly released yet, and I wanted to challenge myself to build something similar using existing tools and APIs.<p>How it works:<p>The user uploads a PDF
The tool extracts the text and feeds it into Google's Gemini Flash language model
Gemini Flash generates a natural, engaging podcast dialogue script based on the key information in the document
This script is then converted to audio using OpenAI's text-to-speech API
The user can listen to the generated "podcast episode" and read along with the transcript
I chose to use Gemini Flash for the language model because it's good at writing high-quality prose while being fast and cheap. We use OpenAI's TTS API to then bring the dialogue to life.<p>Under the hood, it's built with Python, FastAPI, Gradio for the web UI, and my own library, promptic, for calling the LLM and getting structured output. The code is open-source and available on GitHub.<p>Apart from the tool's practical utility, I'm hoping this project can serve as a helpful example for others looking to build applications on top of large language models. It demonstrates an end-to-end flow from document intake to language model usage to audio output, with a simple web interface on top.<p>I would love to hear any feedback or ideas from the HN community! I think there's a lot of potential to expand on this concept and make all sorts of written content more accessible and engaging through audio conversion. Let me know what you think :)
Show HN: PDF to Podcast – Convert Any PDF into a Podcast Episode
Hi HN!<p>I'm stoked to share a project I've been working on called PDF to Podcast. It's a free, open-source tool that automatically converts PDF documents into engaging, informative podcast-style audio content using large language models and text-to-speech tech.<p>Inspiration:
The idea for this project came from the NotebookLM demo at Google I/O, where they showcased generating audio dialogue from uploaded PDFs and other sources. However, that audio feature hasn't been publicly released yet, and I wanted to challenge myself to build something similar using existing tools and APIs.<p>How it works:<p>The user uploads a PDF
The tool extracts the text and feeds it into Google's Gemini Flash language model
Gemini Flash generates a natural, engaging podcast dialogue script based on the key information in the document
This script is then converted to audio using OpenAI's text-to-speech API
The user can listen to the generated "podcast episode" and read along with the transcript
I chose to use Gemini Flash for the language model because it's good at writing high-quality prose while being fast and cheap. We use OpenAI's TTS API to then bring the dialogue to life.<p>Under the hood, it's built with Python, FastAPI, Gradio for the web UI, and my own library, promptic, for calling the LLM and getting structured output. The code is open-source and available on GitHub.<p>Apart from the tool's practical utility, I'm hoping this project can serve as a helpful example for others looking to build applications on top of large language models. It demonstrates an end-to-end flow from document intake to language model usage to audio output, with a simple web interface on top.<p>I would love to hear any feedback or ideas from the HN community! I think there's a lot of potential to expand on this concept and make all sorts of written content more accessible and engaging through audio conversion. Let me know what you think :)
Show HN: Restate – Low-latency durable workflows for JavaScript/Java, in Rust
We'd love to share our work with you: Restate, a system for workflows-as-code (durable execution). With SDKs in JS/Java/Kotlin and a lightweight runtime built in Rust/Tokio.<p><a href="https://github.com/restatedev/">https://github.com/restatedev/</a>
<a href="https://restate.dev/" rel="nofollow">https://restate.dev/</a><p>It is free and open, SDKs are MIT-licensed, runtime permissive BSL (basically just the minimal Amazon defense).
We worked on that for a bit over a year. A few points I think are worth mentioning:<p>- Restate's runtime is a single binary, self-contained, no dependencies aside from a durable disk. It contains basically a lightweight integrated version of a durable log, workflow state machine, state storage, etc. That makes it very compact and easy to run both on a laptop and a server.<p>- Restate implements durable execution not only for workflows, but the core building block is durable RPC handlers (or event handler). It adds a few concepts on top of durable execution, like virtual objects (turn RPC handlers into virtual actors), durable communication, and durable promises. Here are more details: <a href="https://restate.dev/programming-model" rel="nofollow">https://restate.dev/programming-model</a><p>- Core design goal for APIs was to keep a familiar style. An app developer should look at Restate examples and say "hey, that looks quite familiar". You can let us know if that worked out.<p>- Basically every operation (handler invocation, step, ...) goes through a consensus layer, for a high degree of resilience and consistency.<p>- The lightweight log-centric architecture gives Restate still good latencies: For example around 50ms roundtrip (invoke to result) for a 3-step durable workflow handler (Restate on EBS with fsync for every step).<p>We'd love to hear what you think of it!
Show HN: Restate – Low-latency durable workflows for JavaScript/Java, in Rust
We'd love to share our work with you: Restate, a system for workflows-as-code (durable execution). With SDKs in JS/Java/Kotlin and a lightweight runtime built in Rust/Tokio.<p><a href="https://github.com/restatedev/">https://github.com/restatedev/</a>
<a href="https://restate.dev/" rel="nofollow">https://restate.dev/</a><p>It is free and open, SDKs are MIT-licensed, runtime permissive BSL (basically just the minimal Amazon defense).
We worked on that for a bit over a year. A few points I think are worth mentioning:<p>- Restate's runtime is a single binary, self-contained, no dependencies aside from a durable disk. It contains basically a lightweight integrated version of a durable log, workflow state machine, state storage, etc. That makes it very compact and easy to run both on a laptop and a server.<p>- Restate implements durable execution not only for workflows, but the core building block is durable RPC handlers (or event handler). It adds a few concepts on top of durable execution, like virtual objects (turn RPC handlers into virtual actors), durable communication, and durable promises. Here are more details: <a href="https://restate.dev/programming-model" rel="nofollow">https://restate.dev/programming-model</a><p>- Core design goal for APIs was to keep a familiar style. An app developer should look at Restate examples and say "hey, that looks quite familiar". You can let us know if that worked out.<p>- Basically every operation (handler invocation, step, ...) goes through a consensus layer, for a high degree of resilience and consistency.<p>- The lightweight log-centric architecture gives Restate still good latencies: For example around 50ms roundtrip (invoke to result) for a 3-step durable workflow handler (Restate on EBS with fsync for every step).<p>We'd love to hear what you think of it!
Show HN: Restate – Low-latency durable workflows for JavaScript/Java, in Rust
We'd love to share our work with you: Restate, a system for workflows-as-code (durable execution). With SDKs in JS/Java/Kotlin and a lightweight runtime built in Rust/Tokio.<p><a href="https://github.com/restatedev/">https://github.com/restatedev/</a>
<a href="https://restate.dev/" rel="nofollow">https://restate.dev/</a><p>It is free and open, SDKs are MIT-licensed, runtime permissive BSL (basically just the minimal Amazon defense).
We worked on that for a bit over a year. A few points I think are worth mentioning:<p>- Restate's runtime is a single binary, self-contained, no dependencies aside from a durable disk. It contains basically a lightweight integrated version of a durable log, workflow state machine, state storage, etc. That makes it very compact and easy to run both on a laptop and a server.<p>- Restate implements durable execution not only for workflows, but the core building block is durable RPC handlers (or event handler). It adds a few concepts on top of durable execution, like virtual objects (turn RPC handlers into virtual actors), durable communication, and durable promises. Here are more details: <a href="https://restate.dev/programming-model" rel="nofollow">https://restate.dev/programming-model</a><p>- Core design goal for APIs was to keep a familiar style. An app developer should look at Restate examples and say "hey, that looks quite familiar". You can let us know if that worked out.<p>- Basically every operation (handler invocation, step, ...) goes through a consensus layer, for a high degree of resilience and consistency.<p>- The lightweight log-centric architecture gives Restate still good latencies: For example around 50ms roundtrip (invoke to result) for a 3-step durable workflow handler (Restate on EBS with fsync for every step).<p>We'd love to hear what you think of it!
Show HN: A keyboard-centric clipboard history app for macOS
Hey HN,<p>As a Software Engineer, I frequently copy and paste content. The fact that macOS clipboard buffer can keep only one item is very frustrating. It’s very annoying to copy → switch app → paste → switch app → copy → switch app → paste, and so on. Losing items I copied hours ago and having to find and copy them again is a waste of time.<p>I tried many third-party clipboard managers like Alfred, Raycast, Paste, Maccy, etc. While great, they didn't fully meet my needs.<p>As a developer, my requirements are:<p>1. Keyboard-centric operation to avoid using the mouse. I don’t want to waste my time moving my hands from the keyboard.<p>2. Display many clipboard history items at once. I don’t like the idea with big tiles, so I can see only 5-7 history items on my 32” monitor.<p>3. Full content preview for each clipboard item.<p>4. Quick search functionality.<p>So, I created ClipBook for my own use. After months of development, daily use, and positive feedback from my colleagues, I decided to share it publicly. It’s free. All data is securely stored on your computer. I’m planning to make it open source as well.<p>If you try it out, please let me know if you found it useful, if you have anything you’d like me to add, or if you have any other feedback.<p>Happy to answer any questions. Much appreciated.
Show HN: A keyboard-centric clipboard history app for macOS
Hey HN,<p>As a Software Engineer, I frequently copy and paste content. The fact that macOS clipboard buffer can keep only one item is very frustrating. It’s very annoying to copy → switch app → paste → switch app → copy → switch app → paste, and so on. Losing items I copied hours ago and having to find and copy them again is a waste of time.<p>I tried many third-party clipboard managers like Alfred, Raycast, Paste, Maccy, etc. While great, they didn't fully meet my needs.<p>As a developer, my requirements are:<p>1. Keyboard-centric operation to avoid using the mouse. I don’t want to waste my time moving my hands from the keyboard.<p>2. Display many clipboard history items at once. I don’t like the idea with big tiles, so I can see only 5-7 history items on my 32” monitor.<p>3. Full content preview for each clipboard item.<p>4. Quick search functionality.<p>So, I created ClipBook for my own use. After months of development, daily use, and positive feedback from my colleagues, I decided to share it publicly. It’s free. All data is securely stored on your computer. I’m planning to make it open source as well.<p>If you try it out, please let me know if you found it useful, if you have anything you’d like me to add, or if you have any other feedback.<p>Happy to answer any questions. Much appreciated.
Show HN: A keyboard-centric clipboard history app for macOS
Hey HN,<p>As a Software Engineer, I frequently copy and paste content. The fact that macOS clipboard buffer can keep only one item is very frustrating. It’s very annoying to copy → switch app → paste → switch app → copy → switch app → paste, and so on. Losing items I copied hours ago and having to find and copy them again is a waste of time.<p>I tried many third-party clipboard managers like Alfred, Raycast, Paste, Maccy, etc. While great, they didn't fully meet my needs.<p>As a developer, my requirements are:<p>1. Keyboard-centric operation to avoid using the mouse. I don’t want to waste my time moving my hands from the keyboard.<p>2. Display many clipboard history items at once. I don’t like the idea with big tiles, so I can see only 5-7 history items on my 32” monitor.<p>3. Full content preview for each clipboard item.<p>4. Quick search functionality.<p>So, I created ClipBook for my own use. After months of development, daily use, and positive feedback from my colleagues, I decided to share it publicly. It’s free. All data is securely stored on your computer. I’m planning to make it open source as well.<p>If you try it out, please let me know if you found it useful, if you have anything you’d like me to add, or if you have any other feedback.<p>Happy to answer any questions. Much appreciated.
Show HN: Revideo – Create Videos with Code
Hey HN! We’re building Revideo (<a href="https://github.com/redotvideo/revideo">https://github.com/redotvideo/revideo</a>), an open source framework for programmatic video editing.<p>Revideo lets you create video templates in Typescript and render them with dynamic inputs through an API. It also comes with a <Player /> component that lets you preview your projects in the browser and integrate video editing functionality into web apps.<p>The project is useful for anyone who wants to build apps that automate certain video editing tasks. A lot of companies in the space build their own custom stack for this, like Opus (<a href="https://www.opus.pro/" rel="nofollow">https://www.opus.pro/</a>), which automatically creates highlight videos from podcasts, or Clueso (<a href="https://www.clueso.io/">https://www.clueso.io/</a>), which lets you create stutter-free product walkthroughs with AI voiceovers.<p>Revideo is based on the HTML Canvas API and is forked from Motion Canvas (<a href="https://github.com/motion-canvas/motion-canvas">https://github.com/motion-canvas/motion-canvas</a>), a tool that lets you create canvas animations. While Motion Canvas is intended by its maintainer to exclusively be a standalone application [1], we have turned Revideo into a library that developers can integrate into their apps, while specifically focusing on video use cases. To support this, we have, among other things, added the ability to do headless rendering, made video rendering much faster and added support for syncing and exporting audio.<p>We’re excited about programmatic video editing because of the possibility to automate content creation with AI. One of our users is building StoriesByAngris (<a href="https://storiesbyangris.com/" rel="nofollow">https://storiesbyangris.com/</a>), which lets you create video-based RPG stories from language prompts. Other users are marketing-tech companies that help their customers generate and A/B test different versions of video ads.<p>We started to work on video tooling because we ourselves explored a bunch of product ideas in the space of AI-based video creation earlier this year. For example, we built apps that automatically create educational short videos and tinkered with apps that let you create memes.<p>While building these products, we were frustrated with the video editing frameworks we used: Moviepy (<a href="https://github.com/Zulko/moviepy">https://github.com/Zulko/moviepy</a>), which we used initially, doesn’t work in the browser, so we’d often have to wait minutes for a video to render just to test our code changes. Remotion (<a href="https://github.com/remotion-dev/remotion">https://github.com/remotion-dev/remotion</a>), which we switched to later, is pretty good, but we didn’t want to rely on it as it is not FOSS (source-available only).<p>We had already followed Motion Canvas for some time and really liked it, so we thought that extending it would get us to something useful much faster than building an animation library from scratch. We initially tried to build Revideo as a set of Motion Canvas plugins, but we soon realized that the changes we were making were too drastic and far too complex to fit into plugins. This is why we ultimately created a fork. We’re unsure if this is the right way to go in the long term, and would prefer to find a way to build Revideo without feeling like we’re dividing the community - if you have experience with this (keeping forks with complex changes in sync with upstream) or other suggestions on how to solve this, we’d love your input.<p>Our current focus is improving the open source project. In the long term, we want to make money by building a rendering service for developers building apps with Revideo.<p>We’d love to hear your feedback and suggestions on what we can improve! You can find our repo at <a href="https://github.com/redotvideo/revideo">https://github.com/redotvideo/revideo</a>, and you can explore example projects at <a href="https://github.com/redotvideo/examples">https://github.com/redotvideo/examples</a><p>[1] “Motion Canvas is not a normal npm package. It's a standalone tool that happens to be distributed via npm.” - <a href="https://github.com/orgs/motion-canvas/discussions/1015">https://github.com/orgs/motion-canvas/discussions/1015</a>
Show HN: Revideo – Create Videos with Code
Hey HN! We’re building Revideo (<a href="https://github.com/redotvideo/revideo">https://github.com/redotvideo/revideo</a>), an open source framework for programmatic video editing.<p>Revideo lets you create video templates in Typescript and render them with dynamic inputs through an API. It also comes with a <Player /> component that lets you preview your projects in the browser and integrate video editing functionality into web apps.<p>The project is useful for anyone who wants to build apps that automate certain video editing tasks. A lot of companies in the space build their own custom stack for this, like Opus (<a href="https://www.opus.pro/" rel="nofollow">https://www.opus.pro/</a>), which automatically creates highlight videos from podcasts, or Clueso (<a href="https://www.clueso.io/">https://www.clueso.io/</a>), which lets you create stutter-free product walkthroughs with AI voiceovers.<p>Revideo is based on the HTML Canvas API and is forked from Motion Canvas (<a href="https://github.com/motion-canvas/motion-canvas">https://github.com/motion-canvas/motion-canvas</a>), a tool that lets you create canvas animations. While Motion Canvas is intended by its maintainer to exclusively be a standalone application [1], we have turned Revideo into a library that developers can integrate into their apps, while specifically focusing on video use cases. To support this, we have, among other things, added the ability to do headless rendering, made video rendering much faster and added support for syncing and exporting audio.<p>We’re excited about programmatic video editing because of the possibility to automate content creation with AI. One of our users is building StoriesByAngris (<a href="https://storiesbyangris.com/" rel="nofollow">https://storiesbyangris.com/</a>), which lets you create video-based RPG stories from language prompts. Other users are marketing-tech companies that help their customers generate and A/B test different versions of video ads.<p>We started to work on video tooling because we ourselves explored a bunch of product ideas in the space of AI-based video creation earlier this year. For example, we built apps that automatically create educational short videos and tinkered with apps that let you create memes.<p>While building these products, we were frustrated with the video editing frameworks we used: Moviepy (<a href="https://github.com/Zulko/moviepy">https://github.com/Zulko/moviepy</a>), which we used initially, doesn’t work in the browser, so we’d often have to wait minutes for a video to render just to test our code changes. Remotion (<a href="https://github.com/remotion-dev/remotion">https://github.com/remotion-dev/remotion</a>), which we switched to later, is pretty good, but we didn’t want to rely on it as it is not FOSS (source-available only).<p>We had already followed Motion Canvas for some time and really liked it, so we thought that extending it would get us to something useful much faster than building an animation library from scratch. We initially tried to build Revideo as a set of Motion Canvas plugins, but we soon realized that the changes we were making were too drastic and far too complex to fit into plugins. This is why we ultimately created a fork. We’re unsure if this is the right way to go in the long term, and would prefer to find a way to build Revideo without feeling like we’re dividing the community - if you have experience with this (keeping forks with complex changes in sync with upstream) or other suggestions on how to solve this, we’d love your input.<p>Our current focus is improving the open source project. In the long term, we want to make money by building a rendering service for developers building apps with Revideo.<p>We’d love to hear your feedback and suggestions on what we can improve! You can find our repo at <a href="https://github.com/redotvideo/revideo">https://github.com/redotvideo/revideo</a>, and you can explore example projects at <a href="https://github.com/redotvideo/examples">https://github.com/redotvideo/examples</a><p>[1] “Motion Canvas is not a normal npm package. It's a standalone tool that happens to be distributed via npm.” - <a href="https://github.com/orgs/motion-canvas/discussions/1015">https://github.com/orgs/motion-canvas/discussions/1015</a>
Show HN: Revideo – Create Videos with Code
Hey HN! We’re building Revideo (<a href="https://github.com/redotvideo/revideo">https://github.com/redotvideo/revideo</a>), an open source framework for programmatic video editing.<p>Revideo lets you create video templates in Typescript and render them with dynamic inputs through an API. It also comes with a <Player /> component that lets you preview your projects in the browser and integrate video editing functionality into web apps.<p>The project is useful for anyone who wants to build apps that automate certain video editing tasks. A lot of companies in the space build their own custom stack for this, like Opus (<a href="https://www.opus.pro/" rel="nofollow">https://www.opus.pro/</a>), which automatically creates highlight videos from podcasts, or Clueso (<a href="https://www.clueso.io/">https://www.clueso.io/</a>), which lets you create stutter-free product walkthroughs with AI voiceovers.<p>Revideo is based on the HTML Canvas API and is forked from Motion Canvas (<a href="https://github.com/motion-canvas/motion-canvas">https://github.com/motion-canvas/motion-canvas</a>), a tool that lets you create canvas animations. While Motion Canvas is intended by its maintainer to exclusively be a standalone application [1], we have turned Revideo into a library that developers can integrate into their apps, while specifically focusing on video use cases. To support this, we have, among other things, added the ability to do headless rendering, made video rendering much faster and added support for syncing and exporting audio.<p>We’re excited about programmatic video editing because of the possibility to automate content creation with AI. One of our users is building StoriesByAngris (<a href="https://storiesbyangris.com/" rel="nofollow">https://storiesbyangris.com/</a>), which lets you create video-based RPG stories from language prompts. Other users are marketing-tech companies that help their customers generate and A/B test different versions of video ads.<p>We started to work on video tooling because we ourselves explored a bunch of product ideas in the space of AI-based video creation earlier this year. For example, we built apps that automatically create educational short videos and tinkered with apps that let you create memes.<p>While building these products, we were frustrated with the video editing frameworks we used: Moviepy (<a href="https://github.com/Zulko/moviepy">https://github.com/Zulko/moviepy</a>), which we used initially, doesn’t work in the browser, so we’d often have to wait minutes for a video to render just to test our code changes. Remotion (<a href="https://github.com/remotion-dev/remotion">https://github.com/remotion-dev/remotion</a>), which we switched to later, is pretty good, but we didn’t want to rely on it as it is not FOSS (source-available only).<p>We had already followed Motion Canvas for some time and really liked it, so we thought that extending it would get us to something useful much faster than building an animation library from scratch. We initially tried to build Revideo as a set of Motion Canvas plugins, but we soon realized that the changes we were making were too drastic and far too complex to fit into plugins. This is why we ultimately created a fork. We’re unsure if this is the right way to go in the long term, and would prefer to find a way to build Revideo without feeling like we’re dividing the community - if you have experience with this (keeping forks with complex changes in sync with upstream) or other suggestions on how to solve this, we’d love your input.<p>Our current focus is improving the open source project. In the long term, we want to make money by building a rendering service for developers building apps with Revideo.<p>We’d love to hear your feedback and suggestions on what we can improve! You can find our repo at <a href="https://github.com/redotvideo/revideo">https://github.com/redotvideo/revideo</a>, and you can explore example projects at <a href="https://github.com/redotvideo/examples">https://github.com/redotvideo/examples</a><p>[1] “Motion Canvas is not a normal npm package. It's a standalone tool that happens to be distributed via npm.” - <a href="https://github.com/orgs/motion-canvas/discussions/1015">https://github.com/orgs/motion-canvas/discussions/1015</a>
Show HN: Unforget, the note-taking app I always wanted: offline first, encrypted
Hi HN! I created Unforget out of years of frustration with Google Keep and the lack of alternative that met all my needs. I hope you find it useful too!<p>Features include:<p><pre><code> - import from Google Keep
- offline first including search
- sync when online
- own your data and fully encrypted
- Desktop, mobile, web
- lightweight, progressive web app without Electron.js
- markdown support
- programmable with public APIs
- open source [1]
</code></pre>
While I still use org mode for long-form notes with lots of code, Unforget has become my go-to for quickly jotting down ideas and to-do lists after migrating the thousands of notes I had on Google Keep.<p>In addition, I'm thrilled to announce the opening of our software agency, Computing Den [2]. We specialize in helping businesses transition from legacy software, manual workflows, and Excel spreadsheets to modern, automated systems. Please get it touch to discuss how we can help you or if you wish to join our team.<p>[1] <a href="https://github.com/computing-den/unforget">https://github.com/computing-den/unforget</a><p>[2] <a href="https://computing-den.com" rel="nofollow">https://computing-den.com</a>
Show HN: Unforget, the note-taking app I always wanted: offline first, encrypted
Hi HN! I created Unforget out of years of frustration with Google Keep and the lack of alternative that met all my needs. I hope you find it useful too!<p>Features include:<p><pre><code> - import from Google Keep
- offline first including search
- sync when online
- own your data and fully encrypted
- Desktop, mobile, web
- lightweight, progressive web app without Electron.js
- markdown support
- programmable with public APIs
- open source [1]
</code></pre>
While I still use org mode for long-form notes with lots of code, Unforget has become my go-to for quickly jotting down ideas and to-do lists after migrating the thousands of notes I had on Google Keep.<p>In addition, I'm thrilled to announce the opening of our software agency, Computing Den [2]. We specialize in helping businesses transition from legacy software, manual workflows, and Excel spreadsheets to modern, automated systems. Please get it touch to discuss how we can help you or if you wish to join our team.<p>[1] <a href="https://github.com/computing-den/unforget">https://github.com/computing-den/unforget</a><p>[2] <a href="https://computing-den.com" rel="nofollow">https://computing-den.com</a>
Show HN: Unforget, the note-taking app I always wanted: offline first, encrypted
Hi HN! I created Unforget out of years of frustration with Google Keep and the lack of alternative that met all my needs. I hope you find it useful too!<p>Features include:<p><pre><code> - import from Google Keep
- offline first including search
- sync when online
- own your data and fully encrypted
- Desktop, mobile, web
- lightweight, progressive web app without Electron.js
- markdown support
- programmable with public APIs
- open source [1]
</code></pre>
While I still use org mode for long-form notes with lots of code, Unforget has become my go-to for quickly jotting down ideas and to-do lists after migrating the thousands of notes I had on Google Keep.<p>In addition, I'm thrilled to announce the opening of our software agency, Computing Den [2]. We specialize in helping businesses transition from legacy software, manual workflows, and Excel spreadsheets to modern, automated systems. Please get it touch to discuss how we can help you or if you wish to join our team.<p>[1] <a href="https://github.com/computing-den/unforget">https://github.com/computing-den/unforget</a><p>[2] <a href="https://computing-den.com" rel="nofollow">https://computing-den.com</a>
ARC Prize – a $1M+ competition towards open AGI progress
Hey folks! Mike here. Francois Chollet and I are launching ARC Prize, a public competition to beat and open-source the solution to the ARC-AGI eval.<p>ARC-AGI is (to our knowledge) the only eval which measures AGI: a system that can efficiently acquire new skill and solve novel, open-ended problems. Most AI evals measure skill directly vs the acquisition of new skill.<p>Francois created the eval in 2019, SOTA was 20% at inception, SOTA today is only 34%. Humans score 85-100%. 300 teams attempted ARC-AGI last year and several bigger labs have attempted it.<p>While most other skill-based evals have rapidly saturated to human-level, ARC-AGI was designed to resist “memorization” techniques (eg. LLMs)<p>Solving ARC-AGI tasks is quite easy for humans (even children) but impossible for modern AI. You can try ARC-AGI tasks yourself here: <a href="https://arcprize.org/play" rel="nofollow">https://arcprize.org/play</a><p>ARC-AGI consists of 400 public training tasks, 400 public test tasks, and 100 secret test tasks. Every task is novel. SOTA is measured against the secret test set which adds to the robustness of the eval.<p>Solving ARC-AGI tasks requires no world knowledge, no understanding of language. Instead each puzzle requires a small set of “core knowledge priors” (goal directedness, objectness, symmetry, rotation, etc.)<p>At minimum, a solution to ARC-AGI opens up a completely new programming paradigm where programs can perfectly and reliably generalize from an arbitrary set of priors. At maximum, unlocks the tech tree towards AGI.<p>Our goal with this competition is:<p>1. Increase the number of researchers working on frontier AGI research (vs tinkering with LLMs). We need new ideas and the solution is likely to come from an outsider!
2. Establish a popular, objective measure of AGI progress that the public can use to understand how close we are to AGI (or not). Every new SOTA score will be published here: <a href="https://x.com/arcprize" rel="nofollow">https://x.com/arcprize</a>
3. Beat ARC-AGI and learn something new about the nature of intelligence.<p>Happy to answer questions!
Show HN: AI-powered API that instantly obtains website information
I built SiteProfile(siteprofile.io) to simplify how we access detailed website information. The main feature of SiteProfile is that you can obtain all relevant website information with a single, simple API call. With one API call, you can immediately get the following information:<p><pre><code> 1.Real-time Webpage Screenshots
Instantly capture real-time screenshots in both PC or Mobile views.
2. AI-Generated Content:
Generate content based on user prompts and website data, e.g. Describe the core functions of this website.
3.Comprehensive Website Info:
Social Media Links, Contact Info, Basic Details, and Assets, all in one place.
</code></pre>
With just this API, you can instantly create an ai directory or similar website.<p>SiteProfile is currently in the testing phase. During this period, all subscriptions are discounted. There are still a few minor issues, and I’m working hard to fix them.<p>If you try it out, please let me know if you found it useful, if you have anything you’d like me to add, or if you have any other feedback.
And if you have any questions, I’d be happy to answer. Much appreciated.
Show HN: Thread – AI-powered Jupyter Notebook built using React
Hey HN, we're building Thread (<a href="https://thread.dev/" rel="nofollow">https://thread.dev/</a>) an open-source Jupyter Notebook that has a bunch of AI features built in. The easiest way to think of Thread is if the chat interface of OpenAI code interpreter was fused into a Jupyter Notebook development environment where you could still edit code or re-run cells. To check it out, you can see a video demo here: <a href="https://www.youtube.com/watch?v=Jq1_eoO6w-c" rel="nofollow">https://www.youtube.com/watch?v=Jq1_eoO6w-c</a><p>We initially got the idea when building Vizly (<a href="https://vizly.fyi/">https://vizly.fyi/</a>) a tool that lets non-technical users ask questions from their data. While Vizly is powerful at performing data transformations, as engineers, we often felt that natural language didn't give us enough freedom to edit the code that was generated or to explore the data further for ourselves. That is what gave us the inspiration to start Thread.<p>We made Thread a pip package (`pip install thread-dev`) because we wanted to make Thread as easily accessible as possible. While there are a lot of notebooks that improve on the notebook development experience, they are often cloud hosted tools that are hard to access as an individual contributor unless your company has signed an enterprise agreement.<p>With Thread, we are hoping to bring the power of LLMs to the local notebook development environment while blending the editing experience that you can get in a cloud hosted notebook. We have many ideas on the roadmap but instead of building in a vacuum (which we have made the mistake of before) our hope was to get some initial feedback to see if others are as interested in a tool like this as we are.<p>Would love to hear your feedback and see what you think!