Donobu – Mac app turns prompts into deterministic browser tests
Hi HN, we’re Vasusen and Justin, and we’re building Donobu (https://www.donobu.com), a Mac desktop app. It turns prompts like “ensure onboarding works” into reliable browser tests, with optional AI (BYOK). It’s local-first, privacy-focused, and built with insights from our Coursera days—where testing hundreds of features across thousands of pages was a nightmare. Your feedback last year inspired us to double down and enhance Donobu—improving performance, simplifying, and adding seamless integrations with CI/CD. Here's what's new: - 4x smaller install: rewritten in TypeScript,…
In plain words
Donobu is a Mac desktop app that converts natural language prompts into browser tests using Playwright. It generates deterministic, repeatable tests that can be exported and integrated into CI/CD pipelines. The app operates locally and prioritizes privacy, with optional AI capabilities. It's designed for teams testing complex applications across many features and pages, offering features like smart failover handling and support for remote browsers across different locales.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hi HN, we’re Vasusen and Justin, and we’re building Donobu (https://www.donobu.com), a Mac desktop app. It turns prompts like “ensure onboarding works” into reliable browser tests, with optional AI (BYOK). It’s local-first, privacy-focused, and built with insights from our Coursera days—where testing hundreds of features across thousands of pages was a nightmare. Your feedback last year inspired us to double down and enhance Donobu—improving performance, simplifying, and adding seamless integrations with CI/CD. Here's what's new: - 4x smaller install: rewritten in TypeScript, doubles as a local API server at localhost:31000/api. - Deterministic, repeatable Playwright tests exported, with smart failover (no single-selector fragility). - Browserbase (and other remote browsers) support for testing from different locales. - npm package (`npm install donobu`): a superset of Playwright, allowing smart, AI-driven test actions to be added to your existing test suites and enabling integration into CI/CD workflows (e.g., GitHub Actions). - GitHub repo for demo tests created using Donobu, providing practical examples to get you started quickly: https://github.com/donobu-inc/playwright-flows - Full support for major LLM providers (OpenAI, Anthropic, Gemini, AWS Bedrock) plus extended compatibility via Vercel’s AI SDK. `gemini-2.5-experimental` is also supported. - Convenience: download a pre-made llms.txt file from https://www.donobu.com/llms.txt to provide custom context to your preferred LLM chatbot, helping it better understand Donobu usage and generate test cases or API requests. - Added update checks and anonymous, completely disablable, telemetry to notify you about new versions Your prompts, keys, and test data always remain local and private—we don't collect or store them. - Callback URL for integration into your automation systems. What it’s built for: - Crafting and running browser tests, especially on local dev pages. - Testing open-ended, dynamic features like AI-based onboarding flows. - Automating repetitive setup flows What it’s not for: Mass web scraping. Fun things to try: - Asking it to simply test a website (Vibe Testing). - Semantic assertions with visual, subjective tests (e.g., "assert there's a happy theme"). - Prompt in a different language or with a code snippet. Tech Stack: TypeScript, Playwright, Vercel AI SDK, Browserbase. We'd love your feedback, particularly around the desktop app (https://www.donobu.com/download) and npm package. Thanks, Vasusen & Justin
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
Launched alongside, March 2025
the whole month →
Mimic Human Research & Save Findings in AI Knowledge Base
AI · 2025 · sider.ai
