Search

DayZero provides two search modes: structured search with filters and natural language search powered by AI.

Full-Text Search

Search across all entity types in a single query:

bash

curl "https://api.ondayzero.com/api/v1/search?search_term=acme&limit=20" \
  -H "Authorization: Bearer dz_your_token_here" \
  -H "x-business-id: YOUR_BUSINESS_ID"

Parameters

Parameter	Type	Description
`search_term`	string	Search query, 1–200 characters (required)
`entity_type`	string	Limit to a single type (see below); omit to search everything
`date_from` / `date_to`	date (`YYYY-MM-DD`)	Restrict to records created in this range
`limit`	integer	Max results per entity type, 1–100 (10 if omitted). Always pass it explicitly

Search is not cursor paginated: the response has no next_cursor, and the cursor, direction, sort_by and descending parameters of other list endpoints have no effect here. Within each type, results are ordered by relevance (an exact name match first) and then newest first. For types with amounts, a dollar amount such as 1250 or $1,250.00 also matches records whose amount equals it.

Response

Results are grouped by type:

json

{
  "success": true,
  "data": {
    "results": {
      "customers": [ { "id": "…", "name": "Acme Corp", "email": "billing@acme.com" } ],
      "invoices": [ { "id": "…", "number": "INV-1042", "status": "sent", "total": 125000, "customer_id": "…" } ],
      "transactions": [],
      "metadata": { "search_term": "acme", "total_count": 2 }
    }
  }
}

Each item carries its id, its entity_type (singular, for example invoice), and the fields shown for that type; metadata.total_count counts the matches across all types returned. Types with no matches come back as empty arrays.

Searchable Entity Types

`entity_type`	What's searched
`transactions`	Bank transactions (description, counterparty)
`invoices`	Invoice numbers and descriptions
`bills`	Bill numbers and descriptions
`journal_entries`	Entry descriptions
`customers`	Customer names and emails
`vendors`	Vendor names and emails
`ledgers`	Account names and descriptions
`products`	Product names and types
`variants`	Variant names, SKUs and manufacturer SKUs
`projects` / `contracts`	Project names and codes; contract numbers
`emails`	Inbox email subjects, senders and descriptions
`inventory_orders`	Purchase order names, PO numbers and descriptions
`shipments`	Shipment tracking numbers and descriptions

An unknown entity_type returns a 200 with no result groups and metadata.error set to Invalid entity_type: … — check for it rather than assuming an empty result means no matches.

Natural Language Search

Ask a question in plain language. An AI model interprets it and returns a ranked list of suggested results:

bash

curl -X POST "https://api.ondayzero.com/api/v1/search/natural" \
  -H "Authorization: Bearer dz_your_token_here" \
  -H "x-business-id: YOUR_BUSINESS_ID" \
  -H "Content-Type: application/json" \
  -d '{"query": "What did we spend on software last month?"}'

The response has interpreted_query (how the question was understood), results and total. Each result has entity_type, entity_id, title, an optional subtitle and a score between 0 and 1:

json

{
  "success": true,
  "data": {
    "interpreted_query": "transactions categorized as software expenses between 2026-08-01 and 2026-08-31",
    "results": [
      { "entity_type": "transaction", "entity_id": "…", "title": "GitHub", "subtitle": "-$210.00 · 2026-08-04", "score": 0.94 }
    ],
    "total": 1
  }
}

This endpoint returns the model's interpretation, not a database lookup, so do not treat entity_id as a record you can fetch or title as a ledger figure. To retrieve actual records and amounts, use the full-text search above, the entity list endpoints (e.g. GET /api/v1/transactions), or the aggregate Reports.

Searching the documentation

To find the right endpoint rather than your own data, use the docs search at https://www.ondayzero.com/docs/search?q=void+invoice — see Using DayZero from an AI Agent.