
TinyFish provides enterprise infrastructure for AI web agents with a unified platform and single API key.
TinyFish represents a compelling bet on the infrastructure layer for AI agents that interact with the public web. Founded in 2023 by Sudheesh Nair (former Nutanix President), Shuhao Zhang (former Meta), and Keith Zhai (former WSJ), the team combines deep enterprise software, platform engineering, and media-technology expertise.
The company raised a $47 million Series A in August 2025 led by ICONIQ Growth, with participation from USVP, Mango Capital, MongoDB Ventures, ASG, and Sandberg Bernthal Venture Partners. The competitive cohort includes Browserbase, Kernel, Steel, Hyperbrowser, Skyvern, Browser Use, and Firecrawl, but TinyFish differentiates through superior accuracy on hard real-world tasks and a unified API that reduces integration complexity.
TinyFish was founded in 2024 by Shuhao and Gargi, who initially built AgentQL as an AI-powered semantic framework for web automation before expanding into the full TinyFish platform. The company is headquartered in Palo Alto, California.
In February 2026, TinyFish launched a 9-week virtual accelerator in partnership with Mango Capital (led by Robin Vasan), offering a $2M seed funding pool for founders building agentic applications. The program partners with 15+ companies including Google for Startups, Vercel, ElevenLabs, Fireworks.ai, MongoDB, and Composio. TinyFish also makes its Search and Fetch APIs free for all agents, distinguishing it from competitors that charge per query.
TinyFish builds enterprise infrastructure for AI web agents through a unified platform accessible via a single API key. The product suite includes Search, Fetch, Agent, and Browser capabilities designed to automate web-based workflows at scale.
Named customers include Google, DoorDash, ClassPass, Amazon, NextEra Energy Partners, Grubhub, The Zebra, Digital Garage, GetGo, and TestSprite. The platform renders web pages in a full browser and returns clean text content as Markdown or JSON, significantly reducing token usage while improving agent accuracy.
TinyFish provides enterprise infrastructure for AI web agents through a unified platform accessible via a single API key. The platform comprises four core products: a Web Agent that navigates pages, fills forms, authenticates into sites, and returns structured results; a Search API that returns fresh results from the live web as structured JSON; a Fetch API that renders pages in a real browser and returns clean markdown, JSON, or HTML; and a Browser API for session-based browsing.
The Web Agent is benchmarked as the most accurate publicly available web agent, scoring 81.9% on hard tasks on Online-Mind2Web compared to OpenAI Operator at 43.2% and Claude Computer Use at 32.4%. The platform is built on a fully serverless architecture and is used by enterprises including Google, DoorDash, ClassPass, Amazon, and Grubhub.
TinyFish differentiates through superior accuracy on real-world web agent tasks. On Online-Mind2Web hard tasks (300 tasks across 136 live websites), TinyFish scored 81.9% versus OpenAI Operator at 43.2%, Claude Computer Use at 32.4%, and Browser Use at 8.1%. The benchmark uses a human-correlated judge that agrees with humans 85% of the time, avoiding the easy-task bias of WebVoyager.
The platform offers a unified API for search, fetch, browser sessions, and autonomous workflows rather than stitched together point solutions. This consolidation reduces integration complexity and token waste. TinyFish also pioneered making search and fetch free for all agents in May 2026, removing a common cost barrier for agent developers.
TinyFish pioneered making search and fetch free for all agents in May 2026, removing a common cost barrier for agent developers and lowering the marginal cost of web agent operations. This freemium approach aims to drive adoption and ecosystem growth before monetizing advanced autonomous workflows.
Enterprise pricing is likely structured around usage tiers for browser sessions, agent compute, and API call volume, though specific public rate cards were not disclosed at the time of research. The $47 million Series A suggests the company is prioritizing market expansion over near-term monetization.