Browserbase and Parallel: the right search for every agent

TL;DR: We are happy to partner with Parallel to add their Fast and Advanced Search to the Browserbase Search API, enabling every agent to efficiently search the web before acting on it.

Browserbase powers leading AI companies and agents — from AI personal assistants working alongside users to agents completing complex long-horizon tasks in the background.

Across these agent use cases, we’ve learned that effective web automation starts before an agent opens a browser. They first need to identify the right pages and gather current context beyond what their underlying large language models already know. Search provides that starting point; Browserbase’s browser infrastructure then enables them to navigate the web and take action.

Different agents, however, need different search experiences. That’s why we’re excited to partner with Parallel Web Systems to make their search the default provider for the Browserbase Search API and Browserbase Agents. Together, we give every agent, from real-time assistants to long-running autonomous agents, the right way to find where to begin and the infrastructure to complete the work.

Partnering with Parallel

Parallel builds web search for AI agents: fast, token-efficient, and thorough at AI scale. More than 100,000 developers build on Parallel, including teams at Granola, Harvey, Notion, Replit, LangChain, and Opendoor.

Parallel’s search modes let developers trade latency for depth, depending on how long their agent can wait. That maps directly to the two kinds of agents we see on Browserbase: assistants working alongside a user in real time, and autonomous agents running long tasks in the background.

Fast returns results in under a second, so an interactive agent can search mid-conversation without making the user wait. Advanced goes deeper for multi-step research, where finding the right source matters more than saving a few seconds.

By combining Parallel’s search with our browser infrastructure, agents can find the right page and start acting on it in a few lines of code, whether they answer in seconds or run for hours.

A single API, two search modes

Parallel Advanced is now the default search of the Browserbase Search API, a simple request gives you the highest-accuracy results at only $5 for 1,000 requests:

curl -X POST <https://api.browserbase.com/v1/search> \
  -H "Content-Type: application/json" \
  -H "x-bb-api-key: $BROWSERBASE_API_KEY" \
  -d '{
    "query": "a web browser for ai agents",
    "numResults": 10
  }'

Result

{
  "requestId": "string",
  "query": "a web browser for ai agents",
  "results": [
    {
      "id": "string",
      "url": "string",
      "title": "string",
      "image": "Image URL if available",
      "favicon": "Image URL if available",
    }
  ]
}

You can change the search mode to match your agent’s needs, for example, for an interactive AI assistant providing real-time answers:

// get search results in ~800ms
const response = await bb.search.web({
  query: "best ramen in san francisco",
  provider: "parallel",
  mode: "fast", // switch to Fast mode for interactive use cases
  numResults: 5,
});

To act on the web, your agent needs to know where to start

Search gives the agent that starting point. Browserbase Search returns the pages that match your agent's goal, and a Browserbase Browser takes over from there, with the same API key and the same bill. For example:

Business Verification: The search retrieves information from official company registries, such as SEC EDGAR, allowing the browser to look up and display the relevant filings.

Government Forms: The search locates the appropriate forms on official agency websites, enabling the browser to fill them out or download the PDF version.

Shopping and Procurement: The search identifies products across various retailers or supplier websites, allowing the browser to check stock availability, compare options, and complete the checkout process.

Getting started

Parallel is now the default Browserbase Search, available on all plans with 1,000 free searches per month included. Read more in the docs.

Happy building!