Skip to content
Developers

Search

In Tallyfy Answers, search combines AI understanding with keyword matching to find the most relevant results.

Search allows you to find specific objects within your Tallyfy Answers collections1 using natural language queries or specific terms. The system understands both the meaning behind your query and exact keyword matches.

What makes Tallyfy Answers search different?

Section titled “What makes Tallyfy Answers search different?”

Tallyfy Answers uses hybrid search technology2 that combines:

  • Semantic Understanding - Understands the meaning and context of your search query
  • Keyword Matching - Finds exact terms and phrases in each object’s title, content and snippet
  • Source Filtering - Filter results by source

Here’s how to search:

Terminal window
GET /collections/{collection_name}/search/{query}

Search needs no API key. The query goes in the URL path, so URL-encode it: a space becomes %20.

Optional parameters:

  • source - Limit results to one or more sources. Each name is a capital letter followed by lowercase letters, and several names are joined by commas with no spaces, such as Pro or Pro,Answers. Any other value gets a 400.
  • limit - How many results to return, up to 20. A larger value is treated as 20, and a value below 1 gets a 422. Leave it out to get the default, which is 5 unless Tallyfy has changed it.
  • ai_overview - Set to true to ask for a short AI overview of the top results, written by Claude. It’s written in the background, an organization can have it turned off, and there’s a daily limit, so a response may not include one.
  • user_session_id, org_id and user_id - Optional identifiers recorded with the search. org_id also decides whether that organization’s policy allows an AI overview.

If you leave out source, search uses a default source list set on the server, which is Pro unless Tallyfy has changed it. Objects with any other source won’t appear, so pass source whenever you search your own collection.

Example:

Terminal window
curl "https://answers.tallyfy.com/collections/docs/search/workflow%20automation?source=Documentation"

A search returns 200 OK with these fields:

FieldDescription
resultsMatching objects, best match first
results[].idThe object’s uid
results[].title, url, content, sourceCopied from the object. A root-relative url, such as /docs/guide, gets a trailing slash. A full URL comes back unchanged.
results[].hierarchyA list of the values in the object’s hierarchy, in key order. When there are two or more, the last one is left out.
results[].snippetAn extract of content around your search terms, with each term wrapped in <mark>
results[].scoreThe relevance score from reranking, or null when results weren’t reranked
totalHow many results this search can return, up to 20. If results holds fewer, a larger limit gets more.
queryYour query as you sent it
normalized_queryPresent only when punctuation was removed from your query. It’s what was searched.
ai_overviewfalse when you didn’t ask for an overview, it isn’t allowed, or the daily limit is reached. While one is being written, it’s an object with status set to generating and a cache_key. When a finished overview is cached, it holds that overview. It can be missing when the query doesn’t suit an overview.
session_idAn id for this search, present when the service recorded it

A query made only of punctuation gets a 400. A collection that doesn’t exist gets a 200 with no results, the same as a search that matched nothing.

  • Natural Language Queries - Ask questions like “How do I automate workflows?”
  • Multi-word Search - Search with full sentences or phrases
  • Source Filtering - Narrow results to specific content sources
  • Related Suggestions - Get recommendations for similar content from the find similar objects endpoint, which needs the API key

Overview

Tallyfy Answers is a hosted search service that matches the JSON objects you send it by meaning…

Answers > Objects

Objects are the JSON records you store in a Tallyfy Answers collection. They’re indexed in…

Answers > Collections

Collections in Tallyfy Answers group similar objects together, like database tables. You can…
  1. Similar to database tables, collections group related JSON objects with auto-detected schemas ↩

  2. Combines vector embeddings for semantic similarity with BM25 algorithm for keyword relevance scoring ↩