We're getting ready to launch an API that is entirely focused on combining agentic reasoning with RAG -- Is there a name for this? I was thinking something like rawr - reasoning agents with rag

I asked openai deep researcher what R1 was and how I could replicate it on a small scale - the results were very good (no hype) - this is the first thing to impress me as much asthe original ChatGPT release. chatgpt.com/share/67a111c7…

Pillar is amazing to work with and this is an awesome opportunity to get funding for a truly innovative idea.

We renamed R2R "Reason to Retrieve" after witnessing the recent takeoff in reasoning capabilities. This is a demo of the basic R2R RAG agent, we also have a more extensive `deep research` mode. We're excited to see what you build w/ the API - R2R remains OSS + self-hostable!

R2R from sciphi.ai is an open-source agentic retrieval system that transforms RAG with multi-step reasoning across your data and the web. ycombinator.com/launches/N4p-r…

Nothing about my day to day feels quite as good as making devs happy

All it took was ~20 min to implement an MCP server for R2R. It's really juicing up my Claude experience. What else can I do to make this tool useful for others (besides publishing and documenting)?

IMO, the best evaluation of current LLM capabilities is seeing the quality of application a high-agency developer can build overnight. Check out this super clean app with novel AI, built in just 20 hours by YC partner Tom Blomfield: recipeninja.ai

Autists had a great run, the AI future belongs to ADHD

Gemini / Claude are both getting so good that I can't really determine which I prefer easily. It's interesting / helpful to have them both review each-others answers and determine the winner.