Skip to content

Document Search

OMD Cleo provides two document search workflows — one for customer documents stored in the OMD Document Hub, and one for the public OMD documentation site.

The document_search workflow provides an interactive semantic search through documents stored in the OMD Document Hub. Documents are embedded into a vector database (ChromaDB), and the agent retrieves the most relevant passages for each user query.

Workflow ID

document_search

Parameters

No specific workflow_data is required. The workflow is interactive: the agent prompts the user for search queries after the session starts.

Example

{
  "user_ids": ["@user:optimizemyday.ai"],
  "workflow_id": "document_search",
  "workflow_data": {},
  "instance": "sandbox",
  "config_id": "16167225",
  "channel": "matrix",
  "language": "en",
  "new_room": true
}

How it works

  1. The agent greets the user and asks for a search query.
  2. The query is embedded and used to retrieve the most relevant document passages from ChromaDB.
  3. The agent presents the results with source references.
  4. The conversation continues — the user can refine their query or ask follow-up questions about retrieved passages.

Document ingestion

Before documents are searchable, they must be ingested into the vector store. This is done via the /documents API endpoint. Documents are fetched from the OMD Document Hub, chunked, and embedded automatically.

The documentation_search workflow enables users to search through the OMD product documentation at docs.optimizemyday.com.

Workflow ID

documentation_search

Parameters

In addition to the common workflow parameters:

Parameter Type Description
workflow_data.query string The initial user query

Example

{
  "user_ids": ["@user:optimizemyday.ai"],
  "workflow_id": "documentation_search",
  "workflow_data": {
    "query": "How do I configure a new territory?"
  },
  "config_id": "16169276",
  "instance": "www",
  "channel": "matrix",
  "language": "en",
  "new_room": true
}

How it works

The agent performs a semantic search across the indexed OMD documentation pages and returns the most relevant content. The conversation stays open for follow-up questions.