Platform summary

Nimble is web search and retrieval for AI agents. One retrieval stack powers two ways to use it:

• Web Search Agents are the managed research layer. Your agent hands off a task and gets back a completed, cited result.

• Web Tools (Search, Answer, Extract, Extraction Templates, Map, Crawl) are composable APIs for agents and backends that own their own reasoning.

Both run on the same foundation: self-learning retrieval, computer-use browsing with JS rendering, source controls, and governed, schema-enforced output.

Web Search Agents

Web Search Agents turn complex web research into a repeatable API call. You send a task in plain language, optionally with a JSON output schema. The agent plans the work, searches and extracts across the live web, browses sites directly when the data isn't in search results, cross-checks sources, and returns a cited answer or structured dataset.

Three modes:

Mode	What it does	Example
Research	Answers complex questions with a cited brief	Competitive intelligence, due diligence, market analysis
Enrichment	Fills gaps in records you already have, with citations	Add funding, headcount, or pricing to a company list
Dataset building	Builds a fresh structured dataset from scratch across the web	Every clinic in a metro that matches given criteria

What makes them different:

• An expert, not a generic index. You configure an agent's skill, goals, sources, effort, and output schema, so it knows what good output looks like for your domain.

• Self-learning. Memory and a private index store useful sources, outputs, and retrieval paths. Recurring tasks get faster, more accurate, and up to 50% cheaper over time.

• Auditable. You get per-claim citations with verbatim excerpts, graded confidence, and the full list of sources consulted.

• Effort levels run from low to max, so you can trade depth for speed and cost per task.

Web Tools (APIs)

Search

Best for: giving agents fast, real-time web search.

• Input: a query, plus optional filters for time range, domains, and content type (such as PDF)

• Output: structured results with titles, URLs, and snippets. Standard depth also returns rich page content for RAG and agent workflows, and full_content adds the complete page.

• Search depth: Lite for fast, high-volume lookups and Standard (the default) for agent and RAG context

• Focus modes route to specialized sources: general, news, coding, academic, shopping, social, geo, and location

• Results reflect what's on the page right now, not a cached snapshot

Answer

Best for: a synthesized, cited answer to a closed question in one call (Search with include_answer).

Extract

Best for: retrieving clean, LLM-ready content from a known URL.

• Output: markdown, HTML, screenshots, headers, or structured fields from a parsing schema

• Handles difficult pages with JS rendering, browser actions (click, scroll, wait), network capture, and stealth access

Extraction Templates

Best for: typed, normalized data from popular sources without writing selectors.

• Pass a template name and parameters and get back production-ready JSON

• Nimble maintains each template around the clock and auto-heals it when the target site changes

• Supports batch requests of up to 1,000 items, plus webhook or S3/GCS delivery

Map and Crawl

Best for: discovering a site's URLs (Map) or retrieving a whole site section in one request (Crawl), for example to index a docs site into a knowledge base.

How agents use them together

A common pattern: Search finds the right sources, Extract or a Template retrieves clean content from them, and your agent reasons over the result and picks the next step. When the task needs multi-step research, citations, and a schema, delegate it to a Web Search Agent.

Access and integrations

• REST API, Python and JavaScript SDKs, CLI

• MCP server, so agents can call Search, Extract, Map, Crawl, and Web Search Agents as tools

• Framework integrations such as LangChain

• Output delivery to Databricks, Snowflake, S3, and your existing stack

Pricing at a glance

From nimbleway.com/pricing:

• Free trial: 5,000 requests per month, no credit card required, access to all APIs

• Web Search Agents (Agent API): priced per request by effort level: Low $0.025, Medium $0.10, High $0.50, X High $2.00, Unlimited custom

• Search API: $1.10 per 1,000 requests (Lite: links and snippets) or $5 per 1,000 requests (Standard: links and page summaries)

• Extract, Crawl, and Map APIs: $1 per 1,000 requests

• Extraction Templates: $3 per 1,000 requests

• Managed data services: Nimble's experts can configure and manage delivery for your custom data needs; contact Nimble's sales team

• Volume discounts: significant volume discounts are available as usage grows, by speaking with Nimble's sales team. See pricing.

Under the hood

Retrieval runs on Nimble's own browsing infrastructure: JS rendering, real-time site browsing, and a residential network for localized results and reliable access. You don't manage proxies, drivers, or unblocking. Nimble picks the right access path for each source automatically.