I got sick of manually copy-pasting every prompt into 3 different windows just to verify the truth. I realized the only way to get real accuracy was to let the models debate & fact-check each other in real-time, in one screen. I couldn't find any platform online that does this and actually works smooth and user-friendly, so I ended up throwing this together just to make my own life easier.
I built an orchestration layer that runs the models in sequence and manages the state and context windows so they can actually debate. The backend routes the conversation to make sure they are skeptical towards each other's opinions, to make sure they catch each other's flaws in thinking. By talking to each other, the models immediately call out each other’s mistakes and when you push a little more, they definitely don't hold back. It currently supports recent models like Claude Sonnet 4.6, Gemini 3.1 Pro and ChatGPT 5.2.
Running this is pretty token-heavy, but the free tier is open so you can test if the approach works for your workflows. Would love to hear your feedback.
Similar to https://news.ycombinator.com/item?id=47507666 …much to learn from each other, I think.
I am working on a similar hobby project, but for local models and human participation: https://github.com/ch4bes/roundtable
Edit: Asked “what year is it” which gets mixed result if using an LLM without tools.
Results were accurate: https://rauno.ai/c/7nEx8PAVMD
Compared to: https://opper.ai/ai-roundtable/questions/what-year-is-it-7a0...
I tried Opper now you sent it, and they're both similar AI roundtable tools indeed, but my main goal for Rauno was that the user actually sees what every model says, in real-time. To see every perspective of every AI model, so it actually becomes a lively discussion. But opper definitely has some great points too!
Would love to see yours when it's done.
HN is a great place to share projects like these. So much to learn from each other. Thanks for sharing yours!
I do know that coffee comes from East Africa / Ethiopia, and that there are many varieties there due to this.
Heirloom varieties tend to be more “wild”. Farmers growing in regions outside of Ethiopia tend to select stable varieties to match the region.
I don’t know about the frost protection from deep canyons, but it makes sense.
I opened a bag of freshly roasted heirloom beans from Jimma, Ethiopia this morning, and they were indeed more tart and fruity than the Colombian variety I had earlier in the week. The discussion tracked with my experience, at least.