unfortunately we had to shut down, hence decided to open source the code.
There are few products like this online/offline. Thinking it might be useful to somebody, and if so will continue to develop this baby, hence this post.
What it does: allows you to upload huge amount of docs (tested with 100K docs from arxiv), and ask questions to AI agent.
What's different: it's fully contained out of the box, install in a server and work offline, full privacy. Can handle huge sheets/CSV. It's like RAG, but more like research assistant with multi-step processing.
Love to hear feedback.
Why did you have to shut down? What would you have done differently? What would the future roadmap look like if you had been able to continue to build?
We couldn't find paying customers. Maybe we don't know how to cast the net wide to find or the product itself is use less (though it seem useful to me personally). It basically started on wrong footing, we should have started with product market fit first and then build it, we went the other way round :-(
This looks like an interesting tool… if it had helm charts for all of the components I think it would be easier to deploy…
unfortunately we had to shut down, hence decided to open source the code. There are few products like this online/offline. Thinking it might be useful to somebody, and if so will continue to develop this baby, hence this post. What it does: allows you to upload huge amount of docs (tested with 100K docs from arxiv), and ask questions to AI agent. What's different: it's fully contained out of the box, install in a server and work offline, full privacy. Can handle huge sheets/CSV. It's like RAG, but more like research assistant with multi-step processing. Love to hear feedback.
https://news.ycombinator.com/item?id=47228483
Why did you have to shut down? What would you have done differently? What would the future roadmap look like if you had been able to continue to build?
We couldn't find paying customers. Maybe we don't know how to cast the net wide to find or the product itself is use less (though it seem useful to me personally). It basically started on wrong footing, we should have started with product market fit first and then build it, we went the other way round :-(