Browser-based file workspace
Work with audio, video, images and documents in your browser
Cowslator brings local transcription and file tools into one workspace. Add files, describe the task, review the result and download the output. Processing runs on your device where supported; the browser, available memory and selected model affect what can run.
Start with a task
For a recording, open the workspace, choose a transcription model and export a text or subtitle file. For PDFs and images, attach the file and request a supported operation. For documents, extract text, create a local index and ask questions against retrieved passages.
What happens on your device
Selected files are processed in the browser by the tools and models used for a task. Model and runtime assets may need to be downloaded before first use, and browser storage can affect later availability. Read the privacy and storage details before working with sensitive material.
Local processing does not make every task available offline immediately: first-time downloads require a network connection, and speed depends on the device. Cowslator uses browser runtimes such as WebAssembly and, when available, WebGPU.
Check formats and limitations
Available operations depend on the input format and browser support. Document conversion preserves text and data rather than complex page layout; OCR, transcription and larger models can take time and memory.
Supported input formats · Export formats · Browser compatibility · Frequently asked questions
Explore the project
How runtime selection adapts to different devices, what local-first AI does and does not promise, and notes on browser AI engineering.
Cowslator is an independent project by Lochs Studios.