The Unified Data Layer for AI

Power your AI with reliable web data

  • One API for web scraping, search, social media data + more
  • Smart routing across providers for higher success rates & reliability
  • Replace multiple subscriptions and simplify your stack
Get your API key freeNo credit card required.

Works with

Claude CodeOpenClawCursorCodexMake.comn8n

Search, scrape, extract, and enrich from one API

Cover the core external-data jobs your agents need without managing a stack of separate vendors.

Search

Search the web

Run web, news, maps, and autocomplete lookups through one API instead of stitching together separate search vendors.

Auto-route or pin Serper, Perplexity, or DataForSEO

Scraping

Scrape websites

Fetch clean page content from static pages, rendered apps, and full crawls without maintaining brittle scraping pipelines.

Supports Firecrawl, SpiderCloud, and native scraping routes

Documents

Extract document data

Turn PDFs and documents into clean text or structured data for downstream agent workflows and automations.

Use doc-to-text, extract-document, and structured extraction

Social data

Pull social and platform data

Access YouTube, TikTok, LinkedIn, Google Reviews, and other social signals through the same balance and auth layer.

Profiles, transcripts, company data, videos, and reviews

Enrichment

Enrich people and companies

Verify emails, enrich domains, and pull company context without adding one more vendor minimum to your stack.

Pin Hunter or enrich through the unified catalog

Exact control

Drop down to native endpoints

Start with broad capabilities, then pin an exact provider endpoint when upstream behavior, pricing, or output fidelity matters.

Use `firecrawl.scrape`, `serper.search`, or `perplexity.search`

Media

Work with transcripts and media

Pull YouTube and TikTok transcripts, video metadata, and media-derived context without a separate integration surface.

Use transcript APIs alongside search, scrape, and extraction

Automation

Run browser and automation helpers

Use screenshots, crawling, and browser-capable utilities without leaving the same API, CLI, and MCP surface.

Mix screenshots, crawl jobs, and code execution in one stack

Choose the right layer for the job

Capabilities give you the fastest path to working external-data workflows. Native endpoints give you exact upstream behavior when you need it. DumplingAI supports both.

Capabilities

Use when reliability matters

Start with a broad capability like search, scraping, or document extraction when you want DumplingAI to improve the experience above the raw vendor layer.

  • Smart routing across providers
  • Normalized inputs and outputs
  • Material DX improvements above raw vendor APIs
  • Higher data quality through DumplingAI-managed infrastructure
  • Cleaner path to fallback and reliability improvements

Native endpoints

Use when you need provider-specific features

Pin an exact provider endpoint when you need the niche features, quirks, or one-off behavior of a specific vendor surface.

  • Access to provider-specific features
  • Support for quirky or niche upstream endpoints
  • Predictable upstream output and pricing model
  • Easier migration from direct-provider usage
  • Still keeps auth, logs, and spend controls in one place

Start with capabilities when you want DumplingAI's reliability and DX improvements. Drop down to native endpoints when you need a specific vendor's quirks or niche features.

AI native interfaces

Start in the CLI, add MCP or HTTP when you need it

Terminal
# Install & authenticate
$ npm install -g dumplingai-cli
$ dumplingai init

# Run a capability (auto-routes to best provider)
$ dumplingai run capability scrape_page \
    --input '{"url": "https://example.com", "format": "markdown"}'

Why teams choose DumplingAI

One API key, one balance

Stop managing separate subscriptions for search, scraping, transcripts, extraction, and enrichment. One key, one bill, one balance.

Smart routing

Let DumplingAI pick the best provider for each capability, or pin native provider endpoints when you want exact control.

See what your agents spend

Every call logged with cost and latency. Request traces, usage controls, and spend visibility in one place.

Works with your stack

API, CLI, MCP Server, Make.com, and n8n. Claude Code, Cursor, Codex, OpenClaw, or your own runtime.

One balance beats stacking vendor minimums

Search, scraping, social data, transcripts, and enrichment often mean paying several providers before your workflow even runs. DumplingAI puts those capabilities behind one balance and one bill.

Typical vendor stack

Search provider subscription
Scraping provider subscription
Social or transcript provider subscription
Enrichment provider subscription

DumplingAI

One API key. One balance. One bill.

Add capabilities as you expand instead of signing up for one more provider every time your agent needs something new.

  • Broad capability access through one account
  • Routing and provider pinning when needed
  • Spend visibility and request traces in one place

Ship your first agent workflow in 30 seconds

Install the CLI, add your API key, done. Free tier included.

$npm install -g dumplingai-cli
Or browse the docs