AskJ

Cited answers over a closed corpus.

An offline research assistant on a curated writings library — scholar replies with receipts, packaged so others can run it too.

See capabilities Build something like this

Desktop + web · Corpus RAG

Trust the source, not a chatty model.

AskJ is built around one library — not “the whole internet.” Each answer is meant to be checked: citations back to the writings, scholar tone, and a package you can redistribute without standing up a cloud LLM. AnyRAG later generalized the same trust model to any folder you bring.

Capabilities

A closed corpus. Cited scholar answers. Local-first packaging.

Corpus, not crawl: Retrieval stays inside a curated writings set. The product is the library plus the questions you can ask of it — not an open-web chatbot.
Scholar answers: Neutral, cited replies intended for study and verification, not unconstrained generation.
Desktop and web: Same research assistant as a desktop app or a local web UI — without a SaaS backend as the source of truth.
Redistribution: Packaged so the corpus and the assistant can be handed to others who need the same closed library, not a shared cloud tenant.

This is an AI Capability Pack build. Related: AnyRAG (any folder, same trust model).

Who it’s for

Single-author or closed libraries that must stay private.

Scholars & students Ask a known corpus and open the citation, not a vibe.
Publishers & estates A packaged assistant over writings you already control.
Teams who later need “any docs” The pattern became AnyRAG when the corpus is yours to choose.

Built local-first

Proof point for private RAG: the writings stay the boundary. Cloud chat is not the product.

Desktop + web · corpus RAG · cited scholar answers — AI-accelerated delivery, human-architected for a closed library.

Need private AI over a known corpus?

The AI Capability Pack is how we ship products like AskJ — AI-accelerated, human-architected, and yours to own.

Email hello@redn.asia Back to REDN