Describe a task in plain English and an AI agent completes it in a cloud browser — web scraping, structured data extraction, form submission, e-commerce and booking, QA and end-to-end testing, and competitor monitoring. For workflows that need to move beyond the browser and work across desktop applications, terminals, tools, and code, use Navigator n2 through the Navigator API.

Yes. Set require_auth for authentication-heavy flows, or run in local mode via the Yutori Local desktop app to reuse your existing logged-in sessions. Because it drives a real cloud browser, dynamic and JavaScript-heavy pages work too.

Those provide browser infrastructure: cloud browsers you drive with your own code (Playwright, Puppeteer, Selenium) and your own model. Yutori includes the intelligence: Navigator n1.5, Yutori’s browser-optimized computer-use model, runs the browser from a natural-language task, so there are no scripts to write and no model to bring. For broader computer-use workflows, Yutori also offers Navigator n2 through the Navigator API, its flagship model for operating full desktop environments through GUI actions, shell commands, tools, and code.

Those orchestrate third-party LLMs with prompting and DOM heuristics. Yutori runs its own state-of-the-art browser-use model, Navigator n1.5, which outperforms models from frontier labs on web-task benchmarks, delivering higher accuracy at lower latency and cost on fully managed infrastructure. Yutori’s model stack also includes Navigator n2 through the Navigator API, which delivers frontier-level accuracy across OSWorld 2.0, OSWorld-Verified, WeaveBench, MyPCBench, and MacAgentBench at substantially lower token pricing than frontier alternatives.

Yes. Provide an output_schema (JSON Schema) to receive structured results, and a webhook_url (Scout, Slack, or Zapier format) for completion notifications. Tasks run asynchronously with a task_id and a live view URL.