Turn a folder of documents into an LLM wiki.
mighty.MD reads your slides, PDFs, and spreadsheets — diagrams and all — then builds a searchable, backlinked knowledge base with a chat that cites every claim.
Features
Your documents, finally answering back.
Extraction, search, a living graph, and chat — every piece points at one outcome: ask a question, get a cited answer.
Cited chat
Ask anything about your vault. The assistant searches, answers, and links every claim back to its source.
Visual extraction
Claude Opus vision reads slides, diagrams, and charts directly — not just the text layer. Nothing gets lost in the page.
Hybrid search
Keyword and semantic retrieval, fused with RRF — so you find the paraphrase a full-text index would miss.
Backlinked entity graph
Entities are extracted and resolved across documents, then wired into a navigable knowledge graph. Orphan-finder included.
Query datasets in chat
Drop in CSV or XLSX and ask in plain English. Text-to-SQL runs the query against your data and returns the numbers.
How it works
From a folder to a knowledge base — live.
Everything you add rides the same path. Watch it move: read, understood, linked, and ready to query.
- Step 1
Upload
Point mighty.MD at a folder — PDFs, slide decks, spreadsheets, docs.
- Step 2
Extract
Vision and text extraction turn every page, slide, and diagram into clean markdown.
- Step 3
Enrich
People, companies, and products are detected and resolved across documents.
- Step 4
Embed
Each chunk is embedded so search can find meaning, not just matching words.
- Step 5
Graph
Entities and wikilinks resolve into a navigable, backlinked knowledge graph.
- Step 6
Overview
An executive overview is generated, so you start with the big picture and drill in.
Stop reading folders. Start asking them.
Spin up a vault, drop in your documents, and let mighty.MD do the reading.