---
type: "post"
id: "introducing-ultrasearch-for-agents-180e7d"
slug: "introducing-ultrasearch-for-agents-180e7d"
title: "Introducing UltraSearch for agents."
description: "Firecrawl has introduced UltraSearch, an agent-facing search experience connected to its Alexandria data providers."
retrieval_nugget: "Firecrawl has introduced UltraSearch, an agent-facing search experience connected to its Alexandria data providers. Firecrawl advertises searchable data on more than 250 million people, 75 million companies, and 150,000 stocks. For teams building agents, the proposition is richer search results without separately wiring every underlying data source."
published_at: "2026-11-17"
updated_at: "2026-11-17"
record_date: "2026-11-17"
date_kind: "published_at"
topics: ["mcp"]
entities: ["Firecrawl","UltraSearch","Alexandria","WorkOS","MCP"]
related_patterns: ["agent-ready-software-exposes-capabilities"]
novelty_label: "new_fact"
what_changed: "Firecrawl introduces UltraSearch, an Alexandria search experience that returns data-backed answers across stock news, people, and companies for AI agents."
prior_refs: []
source_url: "https://alexandria.firecrawl.dev/go/ultrasearch"
source_urls: ["https://alexandria.firecrawl.dev/go/ultrasearch"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:introducing-ultrasearch-for-agents-180e7d"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/introducing-ultrasearch-for-agents-180e7d/","markdown":"https://newruntime.com/posts/introducing-ultrasearch-for-agents-180e7d.md","json":"https://newruntime.com/posts/introducing-ultrasearch-for-agents-180e7d.json"}
markdown_url: "https://newruntime.com/posts/introducing-ultrasearch-for-agents-180e7d.md"
---

# Introducing UltraSearch for agents.

## Retrieval answer

Firecrawl has introduced UltraSearch, an agent-facing search experience connected to its Alexandria data providers. Firecrawl advertises searchable data on more than 250 million people, 75 million companies, and 150,000 stocks. For teams building agents, the proposition is richer search results without separately wiring every underlying data source.

## Summary
Firecrawl has introduced UltraSearch, an agent-facing search experience connected to its Alexandria data providers. Firecrawl advertises searchable data on more than 250 million people, 75 million companies, and 150,000 stocks. For teams building agents, the proposition is richer search results without separately wiring every underlying data source.

## What happened
Firecrawl’s UltraSearch page describes the product as “answers in your Firecrawl search.” It presents a search interface where a user can select an example and an AI before sending a query. The examples on the page are stock news, people, and companies. Firecrawl’s accompanying launch announcement says UltraSearch returns search results with data snippets rather than positioning the product only as a link-finding tool.

The company highlights three advertised data categories and coverage figures. UltraSearch is described as having access to more than 250 million people with job titles, 75 million companies with funding information, and 150,000 stocks with live prices. Those claims indicate the kinds of entity lookups Firecrawl expects agents to perform: finding a person and role, retrieving a company-related funding fact, or obtaining a stock price. The page does not define the fields returned for every category, nor does it show a complete example response.

Firecrawl explicitly addresses AI agents, LLMs, and automated systems. Its page lists Claude, Claude Code, Codex, Cursor, Grok, Gemini, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Amp, Cline, Goose, Devin, Nous Research Hermes, and OpenClaw among the environments or agents shown. The list is an indication of intended setup compatibility, rather than a statement that each named product has a distinct UltraSearch integration or feature set.

For connection, Firecrawl directs automated systems to an onboarding path that matches their setup. If a platform can mint a WorkOS ID-JAG for Firecrawl, the page points to an authentication guide. Otherwise, it tells the system to read a default onboarding skill and connect through a connector, MCP, or the CLI. Firecrawl says that this connection also gives access to Alexandria’s data providers. It recommends a free account where a human can sign up; where an API key cannot be obtained, it says the onboarding skill describes a keyless free tier for search, scraping, and interaction.

The page links to Firecrawl documentation but does not state a subscription price, credit consumption, request quota, rate limit, or the detailed boundaries of either free-access path. It also does not identify the underlying data providers, explain entity matching or ranking, or define the refresh cadence behind its “live prices” description. The launch establishes an advertised product surface and several connection routes, not a measured claim about search quality.

## What the archive adds
New Runtime’s earlier coverage of [Universal Scrape](https://newruntime.com/posts/introducing-universal-scrape-analysis-7c4b35/) on October 8, 2026 described Firecrawl’s move to consolidate selected external data-provider calls behind an existing endpoint. UltraSearch is a new event on the same product direction: it puts agent search and named entity-data categories at the foreground rather than emphasizing scraping consolidation.

## Trend signal
UltraSearch supports the [agent-ready software exposes capabilities](https://newruntime.com/patterns/agent-ready-software-exposes-capabilities/) pattern. Firecrawl is presenting search over particular data categories as a callable capability available through agent-oriented connection paths, including MCP, connectors, and CLI. That is a more composable boundary than asking an agent to navigate a general-purpose website for each lookup. The announcement, however, does not yet show the permissions, auditability, or reliability controls required to judge how safely this capability can be composed in production.

## Why it matters
Many useful agent tasks depend less on generating prose than on obtaining current facts about an external entity. Recruiting, sales research, market monitoring, and internal enrichment workflows may all need a person’s role, a company’s funding context, or a stock-related data point. A single search layer that can expose those categories to an agent could reduce the integration work needed to prototype such workflows.

That value is conditional. Database size is not the same as answer quality for a particular company, person, geography, or industry. Teams still need to establish whether returned records are current, whether entities are resolved correctly, and whether a result carries enough provenance for the decision at hand. UltraSearch may simplify the retrieval surface, but it does not remove the need for verification when an agent’s output informs consequential work.

## What happens next
- **Firecrawl publishes clearer operational details for UltraSearch access or limits.** (3 months, 60%). Watch for: pricing, credits, rate limits, or a documented free-tier boundary.
- **Builders test UltraSearch in agent workflows that need entity lookup rather than general web search.** (6 months, 65%). Watch for: examples using MCP, CLI, or connectors with people, company, or stock queries.

## Limits and uncertainty
Firecrawl’s announcement gives headline coverage counts but does not disclose the methodology behind them, the underlying providers, update schedules, or geographic and sector coverage. It does not publish independent search-quality benchmarks, latency figures, uptime commitments, or detailed pricing and quota information. The description of live stock prices does not specify the update interval, market coverage, or suitability for trading-related use. No independent evaluation appears on the launch page, so teams should validate the completeness, freshness, and provenance of returned records against their own representative queries.

## Skill to build now
Create a small evaluation set before connecting an agent to an external search layer: ten people, ten companies, and ten market-related queries that matter to your workflow. For each, define the required fields, an acceptable freshness window, and a source you can use to check the result. Run the same set through UltraSearch and record missing records, incorrect entity matches, and unverifiable claims. Use the tool for low-consequence enrichment first; add an approval or independent verification step when an answer changes a customer, hiring, financial, or compliance decision.

## Reader takeaway
UltraSearch is Firecrawl’s new agent-oriented search layer for people, company, and stock data, exposed through familiar setup routes including MCP, connectors, and CLI. Its appeal is integration consolidation: an agent may need fewer separate retrieval interfaces. Treat the advertised coverage counts as a reason to test the product, not proof that its answers are sufficiently fresh or accurate for production decisions.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Mcp](https://newruntime.com/topics/mcp.md): Explore the mcp topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)
