Our mission
Our mission is to build local-first and open-source AI systems that convert unstructured sources (e.g., audio, images, web content, and scholarly articles) into structured, searchable datasets, alongside analysis tools to extract and analyse data at scale.
Built as a third-party extension for Zotero reference manager for systematic reviews and text-based research analysis, it turns a library collection into a project workspace for screening, extraction, semantic search, report development, traceable research workflows, and agentic automation with researcher control.
Transcribe Offline is a free, open-source desktop app that runs entirely on your own hardware for both file transcription and true live transcription, including live speaker diarisation and subtitle export—no cloud uploads or per-minute fees. It supports GPU acceleration on MacOS* and Windows*, but still runs on lower-end devices.
Install or download Transcribe Offline v2.1 for:
OpenResearchTools-Engine is a local AI runtime built to be embedded directly into applications, with a small footprint and one consistent stack for chat, vision, embeddings, reranking, transcription/diarisation, and PDF-to-markdown workflows.
Engine Cluster (experimental) lets you split a runtime across multiple GPUs and machines, and can even mix CUDA, Vulkan, and Metal nodes in the same cluster.
PDF Markdown Studio is a local-first desktop app for converting PDFs and images into clean, editable Markdown—powered by OpenResearchTools-Engine.
Use FAST extraction for machine-readable PDFs, switch to VLM conversion for scanned or layout-heavy documents, or let the app fall back automatically; then review and edit the Markdown in a side-by-side workspace before exporting.