Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.circuit.ai/llms.txt

Use this file to discover all available pages before exploring further.

An index is the core data container in Circuit. It holds your documents and makes them searchable by AI agents. When you upload a document to an index, Circuit processes it, extracting content and analyzing structure to enable intelligent search.

​
Creating an index

1

Navigate to indexes

From your Circuit dashboard, go to the indexes section.
2

Create a new index

Click create and provide:
  • Name: a clear, descriptive name (e.g., “Product Technical Manuals” or “2024 Sales Playbook”)
  • Description: optional, but helpful for other users to understand what’s in it
3

Configure access

Share the index with users or groups who need access. Assign Viewer or Admin roles as appropriate.
4

Upload documents

Add documents to begin building the index. See managing documents for supported file types and upload details.

​
How indexes work

When you add a document to an index, Circuit runs a processing pipeline:
  1. Type detection: identifies the document format (PDF, Word, spreadsheet, CAD, etc.)
  2. Content extraction: pulls text, tables, images, and metadata from the document
  3. Indexing: indexes the content so agents can find and retrieve it
This process runs automatically. You can track the status of each document as it’s processed.

​
Index and agent relationship

Agents search indexes to answer questions. When you create an agent, you configure its reference scope: the set of indexes it can access. An agent can search one or many indexes. This design lets you:
  • Give a support agent access to support articles and FAQs
  • Give a sales agent access to product catalogs, pricing, and competitive analysis
  • Give an RFP agent access to everything: technical docs, compliance, and past proposals

​
Organizing indexes

How you structure your indexes affects search quality, manageability, and security. Prefer building fewer, larger indexes rather than splitting content into many small ones. Larger indexes give agents more context to work with and reduce overhead. The main reason to create separate indexes is to enforce access control boundaries: different indexes can be shared with different users, groups, and agents.
Separate content based on who should have access:
  • “Internal Engineering Documentation” (restricted to engineering)
  • “Customer-Facing Materials” (shared with sales and support)
  • “Confidential HR Policies” (restricted to HR and leadership)
Best for: workspaces where access control is a primary concern.
For detailed guidance, see index best practices.