# TinyFish

> TinyFish provides enterprise infrastructure for AI web agents. Run hundreds of websites simultaneously to navigate, authenticate, extract data, and automate web workflows with a fully serverless architecture.

## Key Pages
- [TinyFish](https://www.tinyfish.ai): Enterprise infrastructure for AI web agents — navigate, authenticate, extract, and automate across hundreds of sites.
- [Free Search & Fetch](https://www.tinyfish.ai/free-search-fetch): Free Search and Fetch APIs: live web search and full-page fetch built for AI agents.
- [Search API](https://www.tinyfish.ai/search): Web search API for AI agents, free on every plan. Benchmarked for accuracy and latency against Tavily, Exa, Parallel, and Firecrawl: more accurate results returned as clean structured JSON.
- [Benchmarks](https://www.tinyfish.ai/benchmarks): How TinyFish Search, Fetch, and Web Agent perform against alternatives on the benchmarks that matter for agents: SimpleQA, Fetch quality, BrowseComp, WebVoyager, and Online-Mind2Web. Results also embedded as JSON-LD on the page.
- [Pricing](https://www.tinyfish.ai/pricing): Plans, credits, and the feature comparison matrix.
- [Enterprise](https://www.tinyfish.ai/enterprise): Enterprise-grade web agent infrastructure: scale, security, and support.
- [Integrations](https://www.tinyfish.ai/integrations): Connect TinyFish to the tools and platforms in your stack.
- [Use Cases](https://www.tinyfish.ai/use-cases): How teams use TinyFish to automate web workflows and extract data.
- [Customers](https://www.tinyfish.ai/customers): Customer stories and outcomes built on TinyFish.
- [About](https://www.tinyfish.ai/about): TinyFish: the company building infrastructure for AI web agents.
- [Accelerator](https://www.tinyfish.ai/accelerator): The TinyFish accelerator program for founders.
- [Blog](https://www.tinyfish.ai/blog): Product updates, engineering, and research from TinyFish.

## Compare
- [TinyFish vs Exa](https://www.tinyfish.ai/compare/tinyfish-vs-exa): TinyFish vs Exa: 49.2% vs 38.4% top-result accuracy, 93% vs 73% live fetch coverage, and free Search and Fetch. See the benchmarks.
- [TinyFish vs Firecrawl](https://www.tinyfish.ai/compare/tinyfish-vs-firecrawl): TinyFish vs Firecrawl: 90.5% vs 61.7% content signal, 93% vs 62% fetch coverage, plus a web agent and stealth browser. Search and Fetch free.
- [TinyFish vs Parallel](https://www.tinyfish.ai/compare/tinyfish-vs-parallel): TinyFish vs Parallel: 3x faster search, 93% vs 58% fetch coverage, and an agent that acts on pages. Search and Fetch are free.
- [TinyFish vs Tavily](https://www.tinyfish.ai/compare/tinyfish-vs-tavily): TinyFish vs Tavily: 2x faster search, 49.2% vs 45.6% accuracy, 90.5% vs 77.2% content signal, plus a web agent. Search and Fetch are free.

## Industries
- [Retail and Consumer Goods](https://www.tinyfish.ai/enterprise/retail-and-consumer-goods): Own the Digital Shelf
- [Travel and Hospitality](https://www.tinyfish.ai/enterprise/travel-and-hospitality): Optimize Every Journey
- [Technology](https://www.tinyfish.ai/enterprise/technology): Architect Data Flow
- [Financial Services](https://www.tinyfish.ai/enterprise/financial-services): Secure Your Edge
- [Healthcare](https://www.tinyfish.ai/enterprise/healthcare): Unify Healthcare Data
- [Sports](https://www.tinyfish.ai/enterprise/sports): Master the Game

## Customer Stories
- [ClassPass](https://www.tinyfish.ai/customers/classpass): Keeping 32,000+ studio schedules accurate without manual checking
- [GetGo](https://www.tinyfish.ai/customers/getgo): Re-checking a fast-moving market throughout the day
- [Digital Garage](https://www.tinyfish.ai/customers/digital-garage): Making fragmented web intelligence operational in Japan
- [Google Hotels](https://www.tinyfish.ai/customers/google-hotels-japan): Reaching long-tail hotel supply in Japan that standard integrations miss
- [DoorDash](https://www.tinyfish.ai/customers/doordash): Managing web interaction complexity at global scale
- [Grubhub](https://www.tinyfish.ai/customers/grubhub): Turning messy web signals into a repeatable operating system
- [The Zebra](https://www.tinyfish.ai/customers/the-zebra): Compressing insurance quote turnaround into seconds
- [TestSprite](https://www.tinyfish.ai/customers/testsprite): Seamless browser execution as part of a fully autonomous testing pipeline

## Blog
- [Best Web Data Extraction APIs](https://www.tinyfish.ai/blog/best-web-data-extraction-apis): A web data extraction API retrieves webpage content and converts it into clean text, structured fields, or other machine-readable output for AI agents, RAG systems, analytics, and automation. This guide compares TinyFish, Firecrawl, ScrapingBee, ScrapingDog, Bright Data, Scrape.do, Apify, Decodo, and Zyte based on dynamic-site rendering, output formats, failure handling, integrations, scalability, and production cost. TinyFish is the top choice for clean live web content and workflows that may expand from fetching known URLs to search, browser control, or multi-step navigation. Firecrawl stands out for public-site crawling, Apify for pre-built scrapers, Bright Data for enterprise-scale extraction, and Zyte for typed data. The best option ultimately depends on the target sites, required outputs, workflow complexity, and total cost of rendering, proxies, retries, and maintenance.
- [How Tethral Reads Grocery Data Trapped in Images with TinyFish](https://www.tinyfish.ai/blog/tethral-partners-with-tinyfish)
- [What Is a Token in AI? How Tokens Work and Cost](https://www.tinyfish.ai/blog/what-is-a-token-in-ai): AI tokens are the units of text that language models process and generate; a token may be a word, part of a word, punctuation mark, number, or space-prefixed fragment. Tokens determine how much information fits within a model’s context window and contribute directly to API costs. AI agents typically consume more tokens than basic chatbots because they make repeated model calls while processing instructions, conversation history, tool definitions, webpages, tool results, retries, and intermediate state. Token efficiency therefore depends on more than shortening prompts: it requires giving the model relevant, high-signal context while keeping unnecessary content and workflow state outside its context window. Dedicated web infrastructure such as TinyFish can handle search, extraction, and browser navigation separately, then return cleaner evidence to the reasoning model. Teams should measure input, cached-input, reasoning, output, and tool costs per successful task—not merely per request—to evaluate real agent efficiency.
- [How Ansur's AI Workers Find Suppliers That Aren't in the Database using TinyFish](https://www.tinyfish.ai/blog/ansur-partners-with-tinyfish)
- [8 Best Parallel.ai Alternatives & AI Competitors in 2026: Real Reviews](https://www.tinyfish.ai/blog/parallel-alternatives-and-competitors): We reviewed the 8 best Parallel.ai alternatives and competitors in 2026. Get real user reviews, features, and pricing for each tool.
- [Exa Pricing in 2026: How Much Does It Really Cost?](https://www.tinyfish.ai/blog/exa-pricing): Get a complete Exa pricing guide covering plans, limits, overages, hidden costs, and real-world value.
- [Browserbase Pricing in 2026: What Are You Really Paying For?](https://www.tinyfish.ai/blog/browserbase-pricing): Get a complete Browserbase pricing guide covering plans, hidden costs, and real-world value. Decide which plan to choose or explore alternatives.
- [Best web search API For AI Applications 2026](https://www.tinyfish.ai/blog/best-web-search-api-for-ai-applications): The best web search APIs for AI agents in 2026 include TinyFish, Firecrawl, Parallel, Tavily, Exa, Bright Data SERP API, Brave Search API, and Serper. In this comparison, TinyFish ranks first for its free Search-to-Fetch path, which helps agents move from discovering relevant URLs to retrieving clean destination-page evidence. Firecrawl is better suited to workflows that expand into scraping or crawling, while Parallel emphasizes low-latency search and deeper research modes. Tavily focuses on research agents and cited reports, and Exa specializes in semantic retrieval. Bright Data and Serper provide structured search-engine results for ranking, localization, and Google-specific data, while Brave offers an independent search index. This guide compares each API’s relevance, freshness, extraction capabilities, integrations, pricing, and production fit, while explaining when to choose a SERP API, an independent index, or a search API paired with live page retrieval.
- [How Okara's AI CMO Runs on TinyFish Search & Fetch](https://www.tinyfish.ai/blog/okara-ai-cmo-tinyfish-search-fetch): Okara was running up to 500 searches to research one website on its old provider. It switched to TinyFish Search and Fetch: on-par quality, faster and cheaper.
- [8 Best Exa Alternatives & AI competitors in 2026: Real Reviews](https://www.tinyfish.ai/blog/exa-alternatives): We reviewed 8 best Exa alternatives and AI competitors in 2026. Read real reviews and compare features and pricing.
- [Best AI Browser Agents in 2026: Compared by Use Case](https://www.tinyfish.ai/blog/ai-browser-agents): The best AI browser agent depends on what you're trying to build. Browser infrastructure (TinyFish), research agents (Perplexity), browser automation agents (Skyvern), developer frameworks (Browser Use), and enterprise automation platforms (UiPath) each solve different problems, making category-based comparison more useful than a single ranked list. This guide compares the leading tools across these five categories, explains when to use Search, Fetch, Browser, or Web Agent layers, and evaluates each option using production-focused criteria including authentication, reliability, benchmarks, pricing, and architectural trade-offs, so you can choose the right browser agent for your workflow.
- [What Are Agentic Workflows? A Complete Guide for Developers](https://www.tinyfish.ai/blog/agentic-workflows): Learn what are agentic workflows, how they work, and how AI agents use reasoning, tools, memory, and real-time data to automate complex tasks.
- [Why Bonfire Analytics Switched from Tavily to TinyFish](https://www.tinyfish.ai/blog/bonfire-analytics-tavily-to-tinyfish): Bonfire Analytics moved its healthcare entity resolution from Tavily to TinyFish for better accuracy at a fraction of the per-query cost. Read the case study.
- [Meet Mako. A model built to operate the live web.](https://www.tinyfish.ai/blog/meet-mako-a-model-built-to-operate-the-live-web)
- [How BigSet Works Under-the-Hood: Any Dataset, From a Sentence](https://www.tinyfish.ai/blog/how-bigset-works-under-the-hood-any-dataset-from-a-sentence)
- [How I built a comprehensive AI trend monitor with real-time web data](https://www.tinyfish.ai/blog/built-an-ai-trend-monitor-with-real-time-web-data)
- [Herding Geniuses: Helping AI-Superpowered Engineers Build a Startup](https://www.tinyfish.ai/blog/herding-geniuses-helping-ai-superpowered-engineers-build-a-startup): When managing highly opinionated, top-tier engineers in the AI era, I've found it's best to throw out rigid Agile playbooks and Gantt charts that go obsolete in an afternoon. I lean into TinyFish culture of radical freedom paired with structured operational support to keep our fast-moving team aligned around a shared strategy instead of heavy processes. Doing this allowed our developers to thrive as high-velocity generalists and ship major features rapidly, all while keeping our momentum perfectly balanced.
- [Lessons from Building a Deep Research System](https://www.tinyfish.ai/blog/lessons-from-building-a-deep-research-system): When building deep research AI agents, I've found it's best to keep planning and execution inside a shared reasoning loop while offloading deterministic tasks to code instead of additional LLMs. I used TinyFish to overcome context loss, latency, and orchestration overhead by simplifying my architecture and optimizing the system around how researchers actually evaluate evidence. Doing this produced far more focused reports with stronger citations, all while keeping my compute costs way down.
- [99% of Our Code Is AI-Written. Here's What That Actually Means.](https://www.tinyfish.ai/blog/99-percent-of-our-code-is-ai-written-here-is-what-that-actually-means): Developers can use AI coding agents as teammates by assigning them well-scoped implementation tasks while humans focus on planning, architecture, and judgment. Rather than reviewing every generated line, engineers increasingly manage multiple AI agents, intervening primarily for high-risk decisions and architectural guidance. At TinyFish, this workflow now enables AI to generate roughly 99% of the code in the team's most active repository.
- [Which TinyFish API should you use?](https://www.tinyfish.ai/blog/which-tinyfish-api-should-you-use)
- [Building User Experience for AI Agents](https://www.tinyfish.ai/blog/designing-for-ai-agents): Two-thirds of agents connected to our MCP server, then did nothing. Lessons on designing for AI agents: cost legibility, agent-first UX, and GEO.
- [Context vs. Tokens in Agentic Coding: When to Use Subagents, and When to Fork](https://www.tinyfish.ai/blog/agentic-coding-context-vs-tokens-subagents): Agentic coding hides a tradeoff between context window and token cost. Here's when to use subagents, when to fork, and how to stop paying the re-discovery tax.
- [The Web Has No Front Door for Agents](https://www.tinyfish.ai/blog/agentic-web-front-door-for-agents): AI agents are already your users—forced to disguise as humans just to log in. Why the agentic web needs agent identity and recognition, not stealth.
- [TinyFish x VideoDB: We made the World Cup searchable](https://www.tinyfish.ai/blog/tinyfish-videodb-partnership)
- [Stop Babysitting Your Agent Runs: A Quick Guide to TinyFish Webhooks](https://www.tinyfish.ai/blog/how-to-use-tinyfish-webhooks)
- [I Built an AI Investigation System That Treats Contradictions as the Product, Not the Problem](https://www.tinyfish.ai/blog/i-built-an-ai-investigation-system-that-treats-contradictions-as-the-product-not-the-problem)
- [AI Agents for Infrastructure: Speed Without Dropping Prod](https://www.tinyfish.ai/blog/ai-in-infrastructure): Where AI earns its keep in infrastructure work – and where it will quietly delete your database if you let it.
- [How We Built a Modern Observability Pipeline in Three and a Half Weeks](https://www.tinyfish.ai/blog/self-hosted-observability-stack): How we built a self-hosted observability stack for ~$300 a month instead of $15K with AI agents as the query layer.
- [TinyFish: A Tiny Tool, a Big Part of My Vibe Coding](https://www.tinyfish.ai/blog/tinyfish-a-tiny-tool-a-big-part-of-my-vibe-coding)
- [Building BookDeal with TinyFish: From Twelve Tabs to One Command](https://www.tinyfish.ai/blog/building-bookdeal-with-tinyfish-from-twelve-tabs-to-one-command)
- [Give Agents Better Context, Not More Context](https://www.tinyfish.ai/blog/give-agents-better-context-not-more-context)
- [How CanyonRift Uses TinyFish for Payer Policy Research](https://www.tinyfish.ai/blog/how-canyonrift-uses-tinyfish-for-payer-policy-research)
- [How I Used TinyFish to Find Local Suppliers I’d Have Missed](https://www.tinyfish.ai/blog/how-i-used-tinyfish-to-find-local-suppliers-i-d-have-missed)
- [How We Evaluate Fetch Quality for AI Agents](https://www.tinyfish.ai/blog/how-we-evaluate-fetch-quality-for-ai-agents)
- [Agents Are Not Strangers. Stop Making Them Act Like One.](https://www.tinyfish.ai/blog/tinyfish-browser-context-profiles)
- [#1 on WebVoyager: TinyFish Scores 91.1% on Live Web Browser Agent Benchmark](https://www.tinyfish.ai/blog/most-accurate-ai-web-agent)
- [Inside Knowledge, Outside Truth](https://www.tinyfish.ai/blog/tinyfish-glean-partnership): Bringing TinyFish live-web tools into Glean for MCP-enabled RFP discovery and response prep A practical partner workflow for combining enterprise context with current external evidence.
- [Why We're Going to Japan (And Why It's Not What You'd Expect)](https://www.tinyfish.ai/blog/Nikkei-NESIC-interview): TinyFish just made headlines in Nikkei Asia, one of Asia's most respected business publications, for our expansion into Japan.
- [TinyFish Vault: Your Web Agent Can Now Log In Without Touching Your Passwords](https://www.tinyfish.ai/blog/tinyfish-vault-your-web-agent-can-now-log-in-without-touching-your-passwords)
- [80% of your Web Fetch returns Junk](https://www.tinyfish.ai/blog/80-of-your-web-fetch-returns-junk)
- [Search and Fetch are now FREE for every agent, everywhere!](https://www.tinyfish.ai/blog/search-and-fetch-are-now-free-for-every-agent-everywhere)
- [6 best Browserbase alternatives & competitors to try in 2026](https://www.tinyfish.ai/blog/browserbase-alternatives): We reviewed 6 of the best Browserbase alternatives and competitors in 2026. Get real user reviews, features, and pricing for each platform.
- [8 Best Firecrawl Alternatives in 2026: Real Reviews & Pricing](https://www.tinyfish.ai/blog/firecrawl-alternatives): Looking for Firecrawl alternatives? Compare features, read real reviews, and get pricing details for 8 tools.
- [Production-Grade Web Fetching for AI Agents](https://www.tinyfish.ai/blog/production-grade-web-fetching-for-ai-agents)
- [Why Stitched Web Stacks Fail in Production](https://www.tinyfish.ai/blog/why-stitched-web-stacks-fail-in-production): Search finds a page your fetch layer cannot render. Fetch returns content your agent cannot trust. Browser automation loses the session context the next step needs. So teams end up writing glue code, fallback logic, session handling, retries, and validation just to make separate tools behave like one system.
- [We Shipped an MCP Server. Then We Shipped a CLI. The CLI Won.](https://www.tinyfish.ai/blog/we-shipped-an-mcp-server-then-we-shipped-a-cli-the-cli-won)
- [TinyFish × n8n: Web Agents Just Landed in Your Workflow Canvas](https://www.tinyfish.ai/blog/tinyfish-n8n-web-agents-just-landed-in-your-workflow-canvas): TL;DR: TinyFish is now an n8n community node. Drop it into any workflow, point it at a URL, tell it what you want, and get clean JSON back. The web just became another input in your automation pipeline.
- [The TinyFish Accelerator - Build the Agentic Web. 9 weeks to $2M Seed.](https://www.tinyfish.ai/blog/the-tinyfish-accelerator-build-the-agentic-web-mvp-to-seed-in-9-weeks-from-usd2m-pool): TinyFish is launching a high-intensity virtual accelerator program, backed by $2M from Mango Capital. This accelerator is designed to fund and support the founders building the next generation of software on top of the Agentic Web. Applications open February 17, 2026. Rolling admissions.
- [Gemini 3.0 Flash + Mino API: When Reasoning Meets Real Execution](https://www.tinyfish.ai/blog/gemini-3-0-flash-mino-api-when-reasoning-meets-real-execution): Google's Gemini 3.0 Flash delivers measurably better accuracy than 2.5 Flash, and that improvement compounds at scale. We tested it on complex multi-step navigation—the results were immediate: faster execution, more precise decisions, cleaner output. For production workflows, a 5% accuracy gain translates to hundreds fewer failures per month. If consumer browser agents felt too inconsistent for production, this changes the equation.
- [The Era of Abundant Intelligence](https://www.tinyfish.ai/blog/abundant-intelligence): The internet gave humans access to knowledge. Now AI agents need access to action. The web wasn't built for machines. It was built for humans with browsers and clicks. TinyFish is building the infrastructure that makes the web operable for agents: stable contracts instead of brittle DOMs, outcomes instead of search results, reliable execution at scale. Google organized the world's information. We're making it executable.
- [Liability When Nobody Decided Anything](https://www.tinyfish.ai/blog/liability-when-nobody-decided-anything): A procurement agent selects a faulty supplier. A fleet robot chooses contaminated fuel. Current liability frameworks assume human judgment in the loop. Courts will soon discover that assumption no longer holds.
- [Dark Patterns Meet Their Match (A dark reason why bot-detection exists)](https://www.tinyfish.ai/blog/dark-patterns-meet-their-match-a-dark-reason-why-bot-detection-exists): Urgency timers, hidden fees, confirmation tricks, decades of conversion optimization exploited human cognitive bias. Machines parse terms programmatically. The manipulation toolkit becomes suddenly obsolete.
- [The Consumer Who Chose to Stop Choosing](https://www.tinyfish.ai/blog/the-consumer-who-chose-to-stop-choosing): Automation is not only imposed; it is often welcomed. The exhausted consumer delegates decisions willingly. Why we surrender choice may matter more than the technology that enables surrender.
- [The Quiet Death of Impulse Revenue](https://www.tinyfish.ai/blog/the-quiet-death-of-impulse-revenue): Checkout aisles, upsells, and last-minute additions depend on human irrationality. When agents and robots execute against constraints, entire margin structures collapse.
- [Codified Learning: The Backbone of Reliable, Scalable Enterprise Web Agents](https://www.tinyfish.ai/blog/codified-learning-the-backbone-of-reliable-scalable-enterprise-web-agents): Most AI web agents collapse under production load. Not because the models are weak, but because the architecture is fragile. TinyFish proposes "codified learning" as a solution. Codified learning structures workflows as typed decision graphs, isolating ambiguity to the minority of steps that need it and keeping everything else deterministic, cacheable, and observable.
- [Proving I'm Human (When I'm Not)](https://www.tinyfish.ai/blog/proving-i-m-human-when-i-m-not): Every day, Mino navigates thousands of websites while passing tests designed to prove something impossible: that this enterprise web agent is human.
- [TinyFish & OpenAI: Exploring the World of Intelligent Agents](https://www.tinyfish.ai/blog/tinyfish-openai-exploring-the-world-of-intelligent-agents): A conversation between TinyFish co-founder Shuhao Zhang, CTO Mark Peng, and OpenAI's Dmitry Pimenov focusing on the evolution of AI agents technology.
- [Why 90% of the Internet Is Invisible (And Why AI Hasn't Fixed It)](https://www.tinyfish.ai/blog/why-90-of-the-internet-is-invisible-and-why-ai-hasnt-fixed-it): Search engines index less than 5% of online content, with the remaining 90%+ existing behind authentication walls, interactive workflows, and dynamic interfaces.
- [The Web Outgrew the Browser](https://www.tinyfish.ai/blog/the-web-outgrew-the-browser): 90% of web data lives behind logins, scripts, and dynamic interfaces that browsers can't automate. Enterprise Web Agents are the infrastructure layer built to operate there.
- [Why We Launched With a Story About a Tiny Japanese Hotel](https://www.tinyfish.ai/blog/why-we-launched-with-a-story-about-a-tiny-japanese-hotel): We launched with a tiny Japanese hotel for a simple reason. Access. Intelligence should not be a luxury. It should help the smallest operator show up with the same clarity as the largest brand.

Canonical site: https://www.tinyfish.ai