Asknes
AI Study Assistant with Multi-LLM Chat & Document Q&A
Got my first sale yesterday, and the feedback so far has been genuinely encouraging. One buyer tested it with a few complex PDFs and said the OCR fallback + LLM responses "worked flawlessly." Another picked it up to speed up building their own SaaS and said it's already helping them stay organized before even digging deep into it.
Small wins, but validating — happy to answer questions if you're curious about the architecture (multi-LLM failover, RAG with citations, OCR fallback).
I've been running Asknes (an AI study assistant — multi-LLM chat with document Q&A, source citations, flashcards, quizzes, and spaced repetition) solo for a while now, and I just packaged the entire codebase as a sellable SaaS boilerplate for other builders.
What's inside:
Multi-LLM setup (Groq, Gemini, Claude) with automatic failover
RAG-based chat with real source citations, not hallucinated answers
OCR fallback for scanned/image-based PDFs
Supabase auth + database, Stripe billing wired in
PWA support out of the box
It's the exact production codebase behind the live app at asknes.com, cleaned up and documented for reuse.
Grabbed it here if you want to skip months of setup and start from a working SaaS: https://tudora5.gumroad.com/l/asknes?wanted=true
Happy to answer any questions about the stack or the build.
-
-
1
Thanks a lot for the support and feedback! Really appreciate you taking the time to test it out. Glad to hear the workflow and PDF processing worked smoothly for you
-
-
1
The fact that this is the production code behind a live app is probably the strongest selling point.
Curious whether buyers are mainly paying to skip the initial implementation work, or because starting from a codebase you’ve already validated removes the risk of stitching all those pieces together themselves.
-
1Honestly, a bit of both — but I'd say the bigger value is the second one. Skipping the initial scaffolding saves maybe a week or two, but the real risk buyers avoid is exactly what you said: stitching together multi-LLM failover, RAG with citations, auth, billing, and OCR fallback into something that actually works together reliably in production. I hit a lot of edge cases building this that aren't obvious until you're live with real users — this codebase already survived that.
-
1That makes sense. The production validation sounds like the more meaningful part of the asset.
-
-
-
-
1
Thank you for buying and taking the time to share this! Hearing that it’s helping you stay organized right away means a lot. If you encounter anything you'd like to see improved as you use it more, I'd love to hear your thoughts
-
Hey IH! 👋 I've been building Asknes — an AI study assistant that actually reads and understands the documents you upload, instead of being a generic chatbot. The core idea: upload your course material (PDFs, notes, scanned docs) and get an AI that can chat with it, cite the exact source for every answer, generate flashcards with spaced repetition, build practice quizzes, and summarize entire documents — even long ones (200+ pages). Some things I'm proud of under the hood: - Multi-LLM setup (Groq, Gemini, Claude) with automatic failover, so it keeps working even if one provider hits a rate limit - RAG-based chat with real source citations, not hallucinated answers - OCR fallback for scanned/image-based PDFs - Works as a PWA, so it installs like a native app on your phone It's live and free to try at asknes.com — built and run solo, still actively improving it. Would love feedback from anyone who tries it, especially students or anyone who studies from dense documents regularly. Happy to answer questions about the tech stack or the build process too!
Comment