TinyFish

TinyFish provides infrastructure that enables AI agents and applications to interact with the live web. Its platform combines web search, webpage retrieval, autonomous browser agents, and managed cloud browser sessions behind a unified API and authentication layer. The Search API returns fresh, structured results from the live web rather than relying solely on cached indexes. It is designed for workflows that depend on changing information such as pricing, market data, competitive intelligence, research, and monitoring. The Fetch API renders webpages in a real browser and converts them into clean Markdown, JSON, or HTML for downstream AI systems. Its Web Agent provides higher-level browser automation. Developers can specify a goal in natural language and let the agent navigate websites, fill forms, authenticate, interact with pages, and return structured results. TinyFish currently reports a 89.9% score on the Mind2Web benchmark for its publicly benchmarked web agent. For workflows requiring direct programmatic browser control, TinyFish Browser provides remote native Chromium sessions accessible through WebSocket/CDP. Existing Playwright, Puppeteer, and raw CDP workflows can connect to these sessions. Residential proxy routing and infrastructure-level anti-bot handling are included in the browser-session price. The platform also supports persistent browser state through Profiles and credential handling through Vault. TinyFish's security architecture is designed so that models do not directly receive credential values; credentials can resolve at the browser layer for authorized actions. TinyFish provides SDKs, a CLI, MCP support, and integrations for AI agent frameworks and platforms, allowing developers to add live web access and browser execution without maintaining their own browser infrastructure.