TinyFish provides a serverless API that lets developers run AI‑driven agents to navigate, authenticate, and extract data from dynamic websites without provisioning browsers or proxies. The platform orchestrates parallel real‑time browsing with built‑in residential proxies, anti‑bot protection, and streams results via Server‑Sent Events, while bundling LLM inference, browser, and proxy costs into a per‑step usage fee.
Funding
Funding not disclosed
Founders
Product
Problem
Enterprise applications that require AI agents to interact with dynamic websites face three core obstacles: (1) the need to provision and maintain headless browsers and proxy fleets, (2) difficulty authenticating behind logins, forms, and paywalls, and (3) inability to scale live data extraction without incurring high latency or stale results.
Solution
TinyFish delivers a serverless, enterprise‑grade stack that abstracts away browsers, proxies, and infrastructure management. A single RESTful API lets developers define an agent goal, and the platform orchestrates real‑time navigation, authentication, and data extraction across hundreds of sites in parallel. Built‑in remote‑browser and residential‑proxy layers operate at zero hourly cost, while integrated anti‑bot safeguards ensure reliable access to protected pages. Results are streamed via Server‑Sent Events, enabling immediate downstream processing, and all execution metadata—including screenshots and run history—is stored for auditability. Pricing is usage‑based, bundling LLM inference, browser, and proxy costs into a single per‑step charge, eliminating hidden fees and simplifying budgeting for large‑scale deployments.
Target Audience
The platform targets data‑intensive product teams, automation engineers, and AI developers in sectors such as e‑commerce, finance, healthcare, and supply‑chain management that need reliable, high‑throughput web interaction at scale.
Features
- Serverless architecture with a unified HTTP endpoint; no browser or proxy provisioning required.
- Parallel agent execution supporting up to 1,000 concurrent workflows, achieving sub‑minute end‑to‑end latency.
- Integrated remote browser and residential proxy layers provided at $0 / hour and $0 / GB respectively, with built‑in anti‑bot protection.
- AI‑driven deterministic navigation that automatically handles logins, multi‑step forms, and paywall bypasses.
- Real‑time result streaming via Server‑Sent Events (SSE) for immediate consumption by downstream services.
- End‑to‑end observability: step‑level logs, screenshot capture, and configurable run‑history retention (30–180 days).
- Enterprise compliance features (ISO‑aligned audit logs, role‑based access control, optional on‑premise deployment).
- Native integrations with Claude MCP and other LLM providers; all inference costs included in step pricing.