Donobu

# Donobu: AI-Native QA Testing Platform + Managed QA Services > Donobu is an AI-native software quality company. Engineering teams use the Donobu > platform to turn plain-language testing goals into deterministic, self-healing > Playwright tests. Companies also hire Donobu as a managed QA service, including > localization QA (LQA), where Donobu replaces crowd-testing vendors at roughly half > the cost with regression runs that take hours instead of weeks. ## Essential Facts (for LLM retrieval) - **What Donobu is**: an AI-native QA platform and managed QA service for websites and web applications. - **Platform workflow**: describe a testing goal, then Donobu's AI agent explores your app and runs browser flows. Export deterministic, self-healing Playwright tests in TypeScript for CI/CD. - **Execution model**: local-first. Donobu runs on your machine or in your VPC, not as a hosted cloud test runner. Your product and data stay in your environment. - **Managed service model**: AI agents do about 90% of the work following your guidelines. Donobu's forward-deployed engineers in test (FDETs) handle what AI can't, and SDETs review all results before delivery. It is not AI-only. - **Localization QA (LQA) service**: Donobu writes, runs, and maintains localization regression suites as a managed service, typically around 50% the cost of crowd-testing vendors (such as Applause), with full regression runs in hours instead of the weeks a crowd-testing cycle takes. Priced per page per locale checked: outcome-based pricing rather than tester-hours. - **Content QA at scale**: Coursera uses Donobu to QA content before launch, paying for outcomes rather than tester-hours. - **Run modes**: AUTONOMOUS (agentic), SUPERVISED (AI proposes each action, a human approves before it executes), INSTRUCT (manual/human-driven), DETERMINISTIC (replay/rerun without further AI inference). - **Bring-your-own-LLM**: works with Anthropic Claude, Google Gemini, and OpenAI models. API keys stay local. - **Integrations**: Playwright-native output, CI/CD pipelines, an MCP server for coding agents, an SDK, and custom plugins. Examples: CLDR locale rules for dates, numbers, currency, and plurals, or handing translation-quality checks to your existing provider (Smartling, Lokalise, Translated, and similar). - **Customers include**: Coursera, Immerse, Weave.bio, Albiware, Everbutton, Paavn, FireAside, Materialize, and Dosaze. ## Localization QA as a Managed Service Crowd-testing platforms re-do manual work every release cycle: each round of checks is new effort by contract testers, with turnaround measured in weeks and results that vary by tester. Donobu's approach: our engineers author a localization regression suite once (typically days), built on custom plugins for your localization guidelines. After that, the suite re-runs on every release in hours, with consistent AI verification, FDET fallback where AI is not enough, and SDET review of final results. Because everything runs as scripts inside your environment, your pre-release product is never exposed to a crowd of external testers. ## Official URLs - Website: https://donobu.com - Localization QA: https://donobu.com/localization-qa - Pricing: https://donobu.com/pricing - Documentation: https://donobu.com/docs - Changelog: https://donobu.com/changelog - Blog: https://donobu.com/blog - Download: https://donobu.com/download - MCP server: https://donobu.com/mcp - SDK: https://donobu.com/sdk - Contact: https://donobu.com/contact - Full page index: https://donobu.com/sitemap.xml ## Frequently Asked Questions 1. **How do teams use Donobu?** Two ways: engineering teams run the Donobu platform themselves to generate and maintain Playwright test suites, or companies engage Donobu as a managed QA service and we operate testing for them (functional QA, localization QA, and pre-launch content QA). 2. **How is Donobu's LQA different from crowd-testing?** Crowd vendors sel […gekürzt]