Analysis updated 2026-07-26
Chat privately with local AI models using markdown and image attachments.
Run AI-assisted coding sessions that preview file changes before applying them.
Generate cited research reports with live progress tracking.
Manage multiple AI chats, coding projects, and scheduled prompts from one app.
| ykesx/dai | 0xkinno/neuralvault | 0xmayurrr/ai-contractauditor | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | hard | easy |
| Complexity | 3/5 | 4/5 | 2/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires a separate local server project called suiban to be installed and running alongside the desktop app.
dai is a desktop application for chatting with local AI models, writing code with AI assistance, and running deep research tasks. The name comes from the Japanese word for the display stand used to present a bonsai tree. Everything runs on your own machine, connected to a separate local server project called suiban. The visual design is meant to feel calm and readable, with ink-on-paper styling, serif headings, and monospace labels for small details. The app has two main sections: Chat and Code. Chat handles streaming conversations with a local AI model, including markdown rendering, code copying, and image attachments for vision-capable models. You can adjust how much effort the model spends on a response, and an ultra toggle lets the orchestrator hand tasks off to sub-agents. Deep research runs inside the chat view, showing progress as a percentage with elapsed time and a cancel option, then displays the finished cited report as markdown. The model picker lets you choose between your local models and external providers like Ollama or OpenAI-compatible endpoints, though external models only support chat, not research or agent features. Code is built for session-based agentic coding. You give it a task and a workspace folder, then watch as it produces a plan, shows a file tree, previews file changes, and logs its activity. Destructive actions like shell commands or file edits pause for your approval first, showing exactly what will happen. A bypass toggle exists but is marked as dangerous and stays off by default. Follow-up messages continue in the same session without starting over. The sidebar organizes everything into chats, tasks, projects, scheduled prompts, and recents. The System page shows whether models are loaded and how much video memory each GPU is using, plus server notices. Settings covers connection details, runtime behavior, quantization options, external providers, connectors, skills management, chat imports, and messaging gateways like Telegram and WhatsApp. A first-run guide appears if required files are missing, showing exact install commands and download sizes. Keyboard shortcuts include a command palette and standard send and line-break controls. The full README is longer than what was shown.
A desktop app for chatting with local AI models, writing code with AI assistance, and running deep research tasks. Everything runs on your own machine for privacy.
Mainly TypeScript. The stack also includes TypeScript, Desktop App.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.