ipdart is one API call that retrieves any URL — across datacenter, ISP, residential, and mobile pools, with stealth fingerprints, JS rendering, and CAPTCHA solving included. Failed fetches cost zero credits.
Waitlist members get an API key and starter credits when access opens. No marketing list.
Your agent needs the live web — a product page, a search result, a documentation site, a listing behind a CAPTCHA. Today that takes three vendors stitched together: a proxy pool, a headless browser service, and a CAPTCHA solver. Each bills you whether or not the page came back usable.
ipdart is a single endpoint that handles all three, and only bills when the bytes are real. Pass a URL, get clean markdown (or HTML, or structured JSON) plus a receipt showing which tier ran the request, how long it took, and what it cost. Built for autonomous callers — idempotent by default, predictable error semantics, no human-only dashboards in the loop.
If we can't return a real 2xx body, you don't pay. No "soft success" page that's actually a CAPTCHA wall. Credits never expire.
Each request is routed to the cheapest tier that returns the page. Pin a tier, or let us escalate automatically when a target gets sticky.
Hand-written SDKs in TypeScript, Python, Go, Ruby, Rust, PHP, Java. First-party MCP server for Claude, Cursor, and other agent hosts.
Real concurrent IPs and per-country counts on a public page. Every response includes which tier, which country, and what we charged.
Sign up, load credits, get an ipd_live_… key. Dual-rail Stripe and crypto from day one.
Send a URL with optional tier, render, and output hints. We route, render if needed, retry on failure.
Clean markdown (or HTML, or schema-validated JSON via /v1/extract) plus a receipt: tier, country, latency, cost.
/v1/fetch — URL in, markdown / HTML / readable view out/v1/extract — LLM-backed structured extraction with a JSON schema/v1/sessions — sticky exit IPs for multi-step logged-in flows/v1/diagnose — pre-flight: which tier will work for this URLIf you're hand-rolling proxy rotation in a Cron job, paying enterprise minimums for a fraction of the volume, or wiring three vendors together to get past a CAPTCHA wall — ipdart is for you.
We're not chasing the Fortune 500 unblocker contract. We're built around small-to-medium scraping operators and the LLM-agent constituency that needs the live web on demand. Per-success billing, transparent pool numbers, open-source SDKs, no annual minimums.
Waitlist first, broader access in waves. Members get an API key, starter credits, and a direct line on which countries and features ship next. We'd rather launch tight to a small group than wide to a marketing list.
Their pools are deeper today. Our pricing is built around small per-success credits instead of six-figure annual contracts, our SDKs are open source, and the surface is shaped for agents rather than for ops engineers ticking a vendor checklist. On the comparison pages at launch, we'll say exactly where they win on pool depth and dataset coverage — we don't pretend to be at their scale yet.
A successful fetch is an HTTP 2xx with a body that passes a content-validity check — actual content, not a CAPTCHA interstitial or a Cloudflare challenge. 4xx and 5xx responses are free. Any retry chain that doesn't end in a real 2xx is free. The whole pricing model is built around that line.
Yes. Pass render=on and we run the target through a hardened Playwright pool with a stealth profile. Or leave it on render=auto and we'll only spin a browser when the static fetch comes back missing content.
We escalate tiers and fingerprints first. If a target still demands a CAPTCHA, we pass through to a solver — and you only pay the additional CAPTCHA credit if the fetch ends in a valid 2xx body. If the solver fails, the whole call is free.
Yes, first-party. Drop it into Claude Desktop, Cursor, or any MCP-compatible host and your agent gets fetch, extract, and diagnose as native tools. Same API key, same idempotency.
Yes — supply cookies or use sticky sessions. We're content-neutral on bring-your-own-credential workflows. The acceptable-use policy draws hard lines at illegal content, sanctioned destinations, and abuse — not at logged-in retrieval.
A solo founder shipping a focused product. The gateway, SDKs, and routing logic are already built — we're finishing compliance plumbing, pool integrations, and a counsel-reviewed AUP before opening the doors.
Waitlist members get a key, starter credits, and a direct line to the founder on what to ship next.