Show HN: Warehouse OpenAI requests to your own database https://ift.tt/5IZzVGb

Show HN: Warehouse OpenAI requests to your own database Today we’re launching Velvet, an AI gateway for warehousing OpenAI and Anthropic requests to your PostgreSQL instance. We originally built an AI SQL editor, but realized that customers were using it to monitor their AI requests in production. We had already built an AI request warehousing tool internally to debug our SQL editor and gave some customers access. A few days into testing this idea, our pilot customer launched [1] and we began warehousing 1,500 requests per second. We worked closely with their engineering team in the following weeks, completely re-architecting Velvet for scale and additional features (such as Batch support). Along the way, other companies began seeking out Velvet to get visibility into their own LLM requests. We’re launching our AI gateway as a self-serve product today, but our pilot customers are already warehousing over 3 million requests per week - so the system is stable and performant. What makes Velvet unique is that you own the data in your own database. Also, we’re the first proxy that gives visibility into OpenAI batch calls - so you can observe and monitor async calls that save you money. Some technical notes: - Supports OpenAI and Anthropic endpoints - Data is formatted as JSON and logged to your own PostgreSQL instance (can add support for other databases for paying customers). - You can include queryable metadata in the header, such as user ID, org ID, model ID, and version ID. - Built on Cloudflare workers, which keeps latency minimal (using our caching feature will reduce latency overall) - Built for security + starting process of SOC II soon Why warehouse your requests? - Understand where money is spent. Use custom headers to calculate the cost per customer, model, or service. - Download real request/response data, so you can evaluate new models (e.g., re-running requests with a cheaper mini model) - Monitor time to completion of batch jobs. (e.g., OpenAI says 24 hours, but our customers average 3-4 hours) - Export a subset of example requests for fine-tuning It’s just a 2 line code change to get started. Try a sandbox demoing the logging proxy here: https://ift.tt/rS8qkoh More details in our docs https://ift.tt/LBmR1V3 [1] https://ift.tt/UI9TW5Z https://ift.tt/RXh4pIH August 28, 2024 at 10:21PM

Show HN: Live Subtitles for Any App on Mac https://ift.tt/rfX0qpZ

Show HN: Live Subtitles for Any App on Mac I’m non-native English Speaker and i’m so tired of saying “Say it again” on online calls. Especially if it is Indian or Scottish Accent. And moreover if it is some really serious interview. That’s why i built Subtitly - Live Subtitles for any app on macOS. Few things about subtitly: - Subtitles are in Mac’s notch ⇒ you always look into the camera - They are really fast and reliable (2-3% Word Error Rate) - On-device AI ⇒ Offline & Private - Works in any app ⇒ Watch twitch streams with subs or calls in discord with subs. Here is the demo how it works: https://ift.tt/wsB4vA1 August 27, 2024 at 09:24PM

Show HN: Built a simple Hacker News client with Expo https://ift.tt/A4XDJ3p

Show HN: Built a simple Hacker News client with Expo Pretty easy to start building native apps with expo these days, especially with prior Web and React knowledge. Did this over a few weekends while learning about Mobile development, their docs are a big help too. https://ift.tt/QFkHfui August 27, 2024 at 06:13PM

Show HN: D&D meets Siri – Interactive voice adventure https://ift.tt/4hUPKco

Show HN: D&D meets Siri – Interactive voice adventure Hey HN! I've been building tooling for voice-driven apps over the past few months, as part of a hardware project. Someone suggested adapting the DSL to play Dungeons and Dragons. So, here we are! What is it? An AI-powered, voice-controlled D&D adventure set in the world of Dvorak. Talk to characters, explore locations, and shape the story using your words. Use your microphone to interact with the AI dungeon master. Explore freely – interrupt, ask questions, or take unexpected actions. If you make friends at the tavern, you can also just hang out there and chat. Hint: Talk to the bartender to move the story along. This is an early demo, and I'm eager for your thoughts: Is the concept engaging? What works well, and what doesn't? I've added a feedback form to the webpage in case you want to drop a comment without posting on HN. Thanks for trying out the demo! https://pocket.computer August 23, 2024 at 06:51PM

Show HN: Lock-free concurrent maps for Golang https://ift.tt/S3vbn4E

Show HN: Lock-free concurrent maps for Golang https://ift.tt/ryl7nb0 August 26, 2024 at 03:02AM

Show HN: Aos – Alias on Steroids https://ift.tt/9Tekaol

Show HN: Aos – Alias on Steroids A simple golang cli to run complex command lines in a simpler way https://ift.tt/FDdB2gI August 26, 2024 at 12:54AM

Show HN: High school robotics code/CAD/design binder release https://ift.tt/M2gATu9

Show HN: High school robotics code/CAD/design binder release Hello HN! My name is Patrick, and I am a junior at my High School’s FRC robotics team FRC 341 “Miss Daisy” (yes named after the movie). Every year, during the first weekend in January, a new robotics game is released (no it’s not battlebots). The game could be about launching balls into a goal, climbing monkeybars, or placing cubes on a see-saw. This year we were challenged to build a robot that could shoot orange foam donuts into a goal about 6 feet in the air. Here is a yt video with the game animation https://www.youtube.com/watch?v=9keeDyFxzY4&t . After we received the game we then had six weeks to design, build, program, and field a robot capable of playing the game. And we did pretty well this year! I have attached a ChiefDelphi thread (robotics forum) where we have released our season materials. This being our CAD (3D model of robot), code, and design binder. I encourage you to take a look and leave any questions that you may have. Thanks a lot! https://ift.tt/pBs893S August 25, 2024 at 01:10AM