1作者: tomkit大约 1 个月前原帖
I forked Chromium and went through the plumbing to embed a ReactJS&#x2F;NextJS web app in its Side Panel like OpenAI&#x27;s Atlas browser and Perplexity&#x27;s Comet browser.<p>The Side Panel has access to the browser&#x27;s DOM which you can pass as context to LLMs in your web app. Supports hot reloading for faster development.<p>Hopefully this is helpful for web devs who haven&#x27;t, or don&#x27;t want to, touch C++ and Chromium, but want to build a native AI experience into browsers.
1作者: amywangyx大约 1 个月前原帖
I&#x27;ve worked as a quant trader for 3 years, and I still had no idea how to allocate my own savings. Sounds ridiculous, but market making at work is very different from figuring out &quot;should I put my retirement money in VTI or gold.&quot;<p>So I built the tool I needed. fin2cents has a portfolio sandbox that runs Monte Carlo simulations using historical return&#x2F;vol profile, updated daily.<p>Most tools show you &quot;7% expected return&quot; and call it a day. But I wanted to know: what&#x27;s my realistic downside?<p>Monte Carlo shows the range of outcomes, not just the average — specifically the 25th–75th percentile, so risk is visible before you commit real money.
2作者: lafalce大约 1 个月前原帖
Hey HN! I built Topic2Manim to automate the creation of educational videos like those from 3Blue1Brown.<p>The workflow is simple:<p>1. Give it any topic (e.g., &quot;how ChatGPT works&quot;)<p>2. An LLM generates an educational script divided into scenes<p>3. LLM generates Manim code for each scene<p>4. FFmpeg concatenates everything into a final video<p>Currently working on TTS integration for narration!<p>Would love feedback on the approach and ideas for the TTS integration
1作者: conradbez大约 1 个月前原帖
YouTube takes free uploaded podcasts clips and charges us outrageous premium fees to view.<p>The obvious alternative is to use a podcast app but I like the 10-20min clips and recommendation engine on YouTube. So I built podtoc.com to serve LLM generated podcast clips youtube-feed style.<p>The Tech:<p>- LLM Pipeline: I built a pipeline to extract meaningful clips from long-form content, specifically designed to capture one of the core insights covered in the podcast.<p>- Recommendation Engine: Suggests clips based on previous listening to solve the discovery problem.<p>- The App: A React Native (Web&#x2F;iOS) app featuring a &quot;swipe to next&quot; UI for seamless browsing.<p>If this sounds like a problem you’ve faced, I’d love to hear:<p>1. Which podcasts would you like to see added to the library?<p>2. Any feedback on the UI or bugs you encounter?<p>3. Any questions about the pipeline or suggestions for the recommendation logic? (would love to open source after some cleanup)<p>Check it out: <a href="https:&#x2F;&#x2F;podtoc.com&#x2F;app" rel="nofollow">https:&#x2F;&#x2F;podtoc.com&#x2F;app</a>