# Exa on x402: find sources for a business research brief

Use Exa's direct search and contents endpoints to gather sources, see the documented request prices and plan a useful first research call.

Source: https://aecon.ai/blog/exa-x402
Author: Agentic Economy
Published: 2026-09-07
Updated: 2026-09-07

We looked at Exa's x402 integration to work out how an agent could gather sources for a supplier brief. We checked its marketplace listing against the provider's documentation and brought together the access path, charging rules and a concrete research task, so you can judge where it might fit in your own work.

The useful distinction is between finding pages and paying to read them. Exa documents separate search and content-retrieval endpoints, giving an agent a chance to select relevant sources before buying more material. A consultant could use that sequence to build an evidence-backed supplier comparison.

The [Exa listing linked from our homepage](https://agentic.market/services/api-exa-ai) points to `api.exa.ai`, the provider's own API domain. We follow that direct route below, with an illustrative request and the documented prices.

## Start with a source-finding task

Consider an illustrative research assignment: identify businesses that offer commercial battery maintenance in Western Australia, then prepare a short comparison of their stated services. The first purchase could be a search for relevant source pages.

The agent should be looking for evidence that will populate the comparison. A company homepage might identify a candidate; a service page might establish whether its offer includes inspections, monitoring or repairs. Neither establishes that the company can take a specific job next week.

This makes the purchased contribution clear. Exa supplies material the agent can use in the larger briefing, as described in our explanation of [how agents buy outside services](/blog/paid-agent-tools).

## The request and result

Send JSON to **`POST https://api.exa.ai/search`**. This illustrative body uses documented fields and has not been run as a paid request:

```json
{
  "query": "commercial battery maintenance services Western Australia",
  "type": "auto",
  "numResults": 5
}
```

Exa's [Search reference](https://exa.ai/docs/reference/search) documents a `results` collection with source URLs and titles, plus metadata such as publication dates and authors. Content options can also request page material. For this first call, treat the returned items as candidate sources for the brief.

To read selected URLs afterwards, use **`POST https://api.exa.ai/contents`**. The [Contents reference](https://exa.ai/docs/reference/get-contents) documents retrieving page material from supplied URLs or identifiers, with options for text, highlights and summaries. Keeping selection and retrieval separate gives the agent a chance to discard irrelevant pages before buying more material.

For the illustrative brief, ask the agent to retain the source URL beside each service claim. Have it distinguish “the company says it offers this” from any judgement about suitability. A missing fact should remain a gap in the comparison.

## What the direct x402 route costs

As checked on **7 September 2026**, Exa's [x402 guide](https://exa.ai/docs/reference/x402-guide) publishes these prices:

| Purchase | Documented price |
| --- | --- |
| `auto`, `fast` or `instant` search | 0.007 USDC per request, up to ten results |
| Contents retrieval | 0.001 USDC per page for each requested content type |
| Search result summaries | Additional 0.001 USDC per result |

Other search modes have different prices. x402 pricing is calculated from request parameters upfront, and its search route caps results at ten. These are USDC amounts, not Australian-dollar prices.

The guide documents keyless access to `/search` and `/contents`. An unpaid request returns the payment requirements; an x402 client can authorise the specified payment and retry. Including an API key instead uses Exa's ordinary account-billing flow. The [Agentic Market API listing](https://api.agentic.market/v1/services/search?q=Exa) corroborates the direct endpoints and the basic search/content prices; its wider results also include separate gateway offers.

## Use the first call to improve the brief

Before paying for a larger search, write the columns you want to fill: provider, claimed service, service area, supporting URL and unanswered question. Then inspect whether the candidate pages actually supply those facts.

If the results are mostly product retailers, the next query should make maintenance more specific. If the business already has a suitable source list, begin with content retrieval. If a source requires interactive browsing to expose the information, a [Browserbase session](/blog/browserbase-x402) is a different kind of contribution.

These are proposed research choices, not benchmark findings about Exa's results. We checked documentation and the public listing; we have not completed a paid search or tested relevance for this example.

For a first implementation, open Exa's linked x402 guide and use its client example with your own bounded query. Read the offered price before authorising it. The [x402 explainer](/x402) covers the payment mechanism, while [Tavily's search route](/blog/tavily-x402) provides another documented option for a current question.