Search each document within a second of indexing it

Give your app full-text search and type-ahead suggestions, with results in under 10ms at the 95th percentile. You push documents to the index through the API, while Conjoin hosts it.

Get early accessRead documentation

Add full-text search to your app through the Conjoin API

Conjoin creates an index within seconds of your API call and makes each document that you push to it searchable within one second. You can add documents one at a time as they change, or in batches for a full re-index.

How Search indexes and ranks your documents

Full-text queries

Search scores each matching document by its relevance to the query and returns the results highest score first, in single-digit milliseconds.

Auto-suggest

The suggest endpoint accepts the prefix that a user types and returns matches from the documents in the same index, so your search box can show results before the user finishes typing.

Batch indexing

Send thousands of documents to an index in a single API call, such as a full export from your product database.

Index management

Create, configure, and delete indexes through the API. When you create an index, name the fields that Search matches against and the fields that you can filter on. Each index can hold a million documents or more.

Document lifecycle

Add, update, or remove one document at a time by its document ID. Search applies each change to its query results in less than a second.

Index provisioning

The create call returns once the index is ready for documents, so your code can push the first batch in its next request. Conjoin chooses the size and the sharding of the index.

< 10ms
Query latency

P95 response time for full-text queries

1M+
Documents per index

The number of documents that one index can hold

< 1s
Indexing delay

Time from a write until queries return the document

Build faster with solutions designed for real-world needs

Add search by creating an index over the API

Create the index with its searchable and filterable fields, then add each product to it as a document with its own ID. Conjoin hosts the index, so these two calls are the whole setup.

Impact

The sample creates the Foilmark Apparel catalog index, with the title, the description, and the tags as searchable fields, and adds the storm shell parka to it. Shoppers can then filter results by category, price, or stock, because the sample declares those fields as filterable.

Prompt for a coding agent: Create a searchable catalog that shoppers can filter by category, price, and stock.

Conjoin IDE

Select a file to view its contents

Add type-ahead with one more endpoint

Every index includes auto-suggest. Your search box sends the prefix that the user types, together with the fields to match and the number of suggestions that it needs.

Impact

In the sample, a customer who types 'esim port' into the Westmark help center receives up to six suggestions from the article titles and section headings. Search returns those suggestions from the same index that your full-text queries search, so you will have no second index to keep in sync.

Prompt for a coding agent: Suggest article titles and headings as customers type in the help center search.

Conjoin IDE

Select a file to view its contents

Re-index a policy library in batches of 1,000

Map each policy to a document and send the documents to the index in batches. Conjoin indexes the documents in a batch in parallel and makes all of them searchable a few seconds after the call returns.

Impact

The sample sends the first 1,000 Eaveline policies to the index in one call, with the line of business, the effective date, and the underwriter of each. Your code can re-index a library of 12,000 policies in 12 such calls, compared with 12,000 calls at one document per call.

Prompt for a coding agent: Re-index our policy library in batches of 1,000.

Conjoin IDE

Select a file to view its contents

Rank results with the default relevance settings

Conjoin tunes the analyzer and relevance defaults of Search for common search workloads, so a query returns ranked results from the first document that you push. You can add filters and pagination to the same query call.

Impact

The sample searches the Driftberry index for coastal pine pollen from spring 2026, limited to sequenced samples from the Pacific Northwest, and requests the first page of 20 results. Search ranks those results with its defaults, so the code sets no analyzer or scoring option.

Prompt for a coding agent: Search the spring 2026 pollen samples from the Pacific Northwest, 20 results a page.

Conjoin IDE

Select a file to view its contents

Build your next app on Conjoin

Conjoin will open with a free Build tier, while the pricing page already lists the launch prices for Pro and Business. If you tell us what you're building, we'll talk you through early access.

See pricingGet early access