TinyFish

TinyFish is an AI web agent tool built for teams that need live web actions rather than static scraped data. It turns plain-English goals into browser actions, then navigates websites, logs in, extracts data, and completes workflows through a visual workbench or API. It is aimed more at developers, ops teams, and enterprise workflows than casual users, and that focus is part of its appeal.

Key Features

• Natural language control: Users describe a task in plain English instead of writing selectors or XPath.
• Live execution streaming: Runs can be watched in real time with streaming events and browser previews.
• Stealth and proxy options: TinyFish offers anti-detection browser modes and geo-targeted proxy support for harder sites.
• Parallel agent runs: Plans support from 2 to 50 concurrent agents for higher-volume workflows.
• Workbench, API, and MCP access: Teams can build visually or connect TinyFish to code-based and assistant-driven workflows.

• Works on messy real-world websites: It supports dynamic sites, authenticated systems, and multi-step workflows.
• Built for scale: TinyFish is clearly made for repeated, high-volume web operations, not one-off tinkering.
• Good observability: Screenshots, run history, and live previews make debugging far less annoying.
• Clear cost structure: Browser, proxy, LLM, and anti-bot costs are bundled, which is refreshingly less sneaky than many usage-based tools.

• Not especially beginner-first: The product language and setup lean toward technical or workflow-heavy teams.
• CAPTCHA handling is still limited: TinyFish states CAPTCHAs cannot currently be solved automatically.
• Public integration detail is fairly light: API and MCP support are clear, but the site exposes fewer ready-made app details than some rivals.

Who Uses It

• E-commerce and retail teams: For product matching, pricing checks, and checkout monitoring.
• Travel and hospitality businesses: For hotel availability, rate monitoring, and property data collection.
• Sales and lead generation teams: For structured lead extraction from conference and company websites.
• Developers and operations teams: For browser automation through APIs, async runs, and agent-based workflows.
• Uncommon use cases: Sports scouting across global sources and healthcare portal unification stand out as more unusual but very fitting uses.

Pricing

• Pay-as-you-go: $0.015 per step. Includes 2 concurrent agents, remote browser at $0 per hour, residential proxy at $0 per GB, all LLM costs included, anti-bot protection, 30-day run history, and email support.
• Starter: $15 per month. Includes 1,650 steps per month, 10 concurrent agents, remote browser at $0 per hour, residential proxy at $0 per GB, all LLM inference costs included, anti-bot protection, 30-day run history with observability and screenshots, priority email support, and TinyFish Workbench, API, and MCP integration. Overage is $0.014 per step.
• Pro: $150 per month. Includes 16,500 steps per month, 50 concurrent agents, remote browser at $0 per hour, residential proxy at $0 per GB, all LLM inference costs included, anti-bot protection, 180-day run history with observability and screenshots, priority email plus Slack/Discord community access, and TinyFish Workbench, API, and MCP integration. Overage is $0.012 per step.
• Enterprise: Custom pricing. Includes custom workflows, custom step volume, custom concurrent agents, custom auditability, advanced security and compliance, dedicated support engineer, custom SLA and contracts, and on-premise deployment options.