Local-first by default
Run compatible local models, keep project files nearby, and use offline workflows without mandatory cloud accounts or API keys.
HugstonOne brings local model work, chat, files, retrieval, research, code preview, tools, skills, and agents into one focused workspace. Local mode keeps work on the machine or infrastructure you control. Online mode is optional and explicit.
Local-first AI workspace for Windows x64. Legacy Community Edition 1.0.8.
Run compatible local models, keep project files nearby, and use offline workflows without mandatory cloud accounts or API keys.
Bring documents, code, tables, images, and project material into chat, retrieval, analysis, and isolated workspaces.
Use chat, RAG, code and content previews, reusable skills, controlled tools, and agent workflows from one interface.
Enable network-backed research only when it is needed. Local work remains available when external sources are unavailable.
Choose models and runtime modes, stop generation, manage memory behavior, and preserve work without hidden forced updates.
Support research, writing, coding, learning, private knowledge work, prototypes, structured outputs, and repeatable workflows.
Start in local mode, select a compatible model, open or create a session, and add only the files needed for the task. Turn on retrieval, tools, skills, agents, or online mode deliberately. Results remain visible in the same workspace, including previews and generated project files when those features are used.
The Community Edition is free and available in portable, setup EXE, and MSI formats. It remains a legacy release and is not actively updated. The Enterprise Edition is the current paid product.
The free Community Edition is available as a public legacy release and is not actively updated.
The paid Enterprise Edition is the current edition. Contact the Hugston.com team for payment details and access.
The public guides explain what Hugston.com and HugstonOne do, how their main user workflows operate, the edition differences, privacy expectations, and sensible operating limits. They intentionally do not expose private infrastructure, credentials, or implementation-sensitive details.