Hi fellow devs! I built a dev tool (database abstraction layer) and I'm testing the UI/UX. Is the UI clear enough that you know what to do without reading docs? Would love any feedback (and happy to return the favor), thanks! --> app.clarabase.com
app.clarabase.com
Clarabase
Create no code, no deploy API services in seconds
With AI, everybody will make their own software just like how we all 3D print our own furniture.
New weekend hack: Remember those "Pepsi vs Coke" blind taste test? I built that for LLMs. It’s fun to see how our opinions about which LLM is "better" changes when you remove bias caused by branding/logos/etc. Try it out 👉 llm-blind-test.onrender.com
"Vibe coding" does not mean any coding with AI. If you are vibing, you're not looking at the code. Building something throwaway or a prototype. That is very different from using AI to write code that is reviewed, tested, and maintained over time. x.com/leerob/status/…
Here's an example of how I used Cursor to build Cursor. And where AI models failed and human review was necessary! I wanted to add a /compress command to summarize all messages in a conversation. This is helpful because you can manually decide to reset your context window,…
Model versions and benchmark scores are starting not to map cleanly to how we perceive of improvements between LLM versions. Maybe we need a "perceptual curve" benchmark to measure the leaps humans actually notice (like what the Fletcher-Munson curve does in the audio world)
Making more progress with building my agentic ML engineer project (it creates an ML pipeline and trained model for you given a prompt + data) Have the basic flow working locally, next step is moving it out of local dev into a scalable cloud env!
where do the emdashes even come from. what corpus has that many emdashes
New oss models from OpenAI look promising, can't wait to dive in! openai.com/index/introduc…
Teaching myself google's new MLE-STAR agent by building a tiny app that takes your plain-language prompt & dataset and returns a full ML pipeline + trained model. Learning a lot, can’t wait to share more! 🚀
Just launched: OpenJury 🏛️ A new open source framework for evaluating LLM outputs using a configurable panel of LLM jurors 🤖⚖️ ✅ Post-inference ensembling ✅ Multi-criteria, weighted voting Built for devs, researchers, & AI teams. github.com/robiscoding/op…
I'm curious if dependency trees are getting less bloated these days. Complex libraries (i.e. graphing) are always needed, but for simpler utils, why would anyone install a 100kb pkg when AI can generate exactly what you need?
Does your knowledge of whether or not AI was used to build a product or feature affect your view of the quality of the product/feature?
Does anyone know of an empirical study that concludes that AI speeds up development?
United States 趨勢
- 1. Yamamoto 46.3K posts
- 2. #DWTS 43.1K posts
- 3. Brewers 41.2K posts
- 4. Growth Path 1,604 posts
- 5. #FlyTogether 2,632 posts
- 6. #TexasHockey 3,336 posts
- 7. halsey 8,162 posts
- 8. Ohtani 13.9K posts
- 9. Kreider 1,199 posts
- 10. Young Republicans 72.4K posts
- 11. Jared Butler N/A
- 12. #Dodgers 16.1K posts
- 13. #MakeOffer 10.5K posts
- 14. Domain For Sale 10.8K posts
- 15. Jarry N/A
- 16. Robert 106K posts
- 17. Tokyo Narita N/A
- 18. Haji Wright 1,203 posts
- 19. Will Richard 2,622 posts
- 20. Roldan 2,662 posts
Something went wrong.
Something went wrong.