Skip to main content
Charcoal is an API for intelligent document search. Upload your documents, and Charcoal indexes and searches them using AI, returning synthesized answers with cited excerpts instead of just keyword matches.

​
How it works

  1. Organize documents into namespaces. Namespaces are collections of documents with a shared schema. Think of them like tables in a database.
  2. Upload documents. Documents are JSON objects with an id and arbitrary attributes. Define which attributes are searchable and filterable.
  3. Search with natural language. Describe what you’re looking for with an objective and context. Charcoal returns a synthesized answer backed by specific findings and excerpts from your documents.

​
Key features

  • AI-powered search. Go beyond keyword matching. Describe what you need in natural language and get synthesized, cited results.
  • Structured documents. Documents are schematized JSON with typed attributes, so you can combine natural language search with precise attribute filters.
  • Streaming results. Get real-time progress updates as search executes via server-sent events.
  • Multi-turn search. Search sessions support follow-up messages when the initial query needs clarification or refinement.

​
Next steps

Installation

Install and configure Charcoal.

API Guide

Upload documents and run your first search.

API Reference

Full API endpoint documentation.

Search

Learn about search features, streaming, and multi-turn sessions.