osFoundry / Platform
Knowledge Base
Build a searchable, versioned KB from docs, PDFs, or scraped sources. Indexed automatically.
What it is
osFoundry's Knowledge Base turns a pile of documents into a searchable, versioned reference that lives inside your workspace. You point it at what you already have: upload Markdown, HTML, PDF, DOCX, or CSV files, bring in structured exports from Notion, Confluence, GitBook, or WordPress, or scrape an external page by URL. Each source is parsed into clean articles and organized into categories, so a folder of raw files becomes browsable documentation rather than a download you forget about.\n\nIndexing happens automatically. When content is ingested it is chunked and embedded, then stored so it can be retrieved by meaning rather than exact keywords. Search combines semantic (vector) matching with keyword lookup and reranks the results, which is how it surfaces the right paragraph even when your question uses different words than the document. Articles are versioned: every edit keeps a revision history, so you can see how a definition or procedure changed over time and roll back if an update was wrong. Because this all runs from your osFoundry workspace, the KB isn't a separate SaaS silo, it's wired into the same place you chat and run agents.\n\nThat wiring is the point. The indexed Knowledge Base is available as retrieval context for your models, so chat and agents answer from your actual documents and can point back to the source article instead of improvising. osFoundry is bring-your-own-key and local-first: you connect your own model API keys with no reseller markup, and on desktop your content can stay on your device. One base can serve support answers, internal runbooks, product docs, or research notes, all searchable, all versioned, all in one workspace.
Key capabilities
- Import from many sources: Markdown, HTML, PDF, DOCX, and CSV files, plus Notion, Confluence, GitBook, and WordPress exports, or scrape content directly from a URL
- Automatic indexing on ingest: documents are chunked and embedded so the whole base becomes searchable without a manual reindex step
- Versioned articles: every edit keeps a revision history you can review, so you can see how an answer changed and restore a prior version
- Hybrid search with reranking: semantic (vector) matching combined with keyword retrieval finds the right passage even when wording differs
- Feeds your chat and agents: the same indexed KB is available as retrieval context, so answers cite your documents instead of guessing
Use cases
- Support lead: import existing help docs and past Confluence articles into one KB so agents answer customer questions from approved, versioned content instead of guessing
- Solo founder / operator: scrape and upload scattered PDFs, runbooks, and notes into a single searchable base that your own AI chat can cite, using your own API keys
- Research or product team: keep a versioned reference of specs and findings, review how a definition changed over time, and feed it to agents as retrieval context
Frequently asked questions
What file types and sources can I build a knowledge base from?
You can upload Markdown, HTML, PDF, DOCX, and CSV files, import structured exports from Notion, Confluence, GitBook, and WordPress, or scrape content from a URL. Each source is parsed into articles and sorted into categories, and everything is indexed for search once it's in.
Does the knowledge base index my documents automatically, or do I have to run indexing myself?
It indexes automatically. When you add content it is chunked and embedded on ingest, so the base becomes searchable right away without a separate manual reindex step. Search then uses semantic matching plus keyword retrieval with reranking to find the relevant passage.
Can I see the history of an article or undo a bad edit?
Yes. Articles are versioned, so every edit is kept as a revision. You can review how an article changed over time and restore an earlier version, which matters when a doc is a shared source of truth that several people update.