aayushch / laya
A local-first AI command center that transforms multi-platform notifications into pre-researched, ready-to-approve Action Cards using specialized AI personas.
星标趋势
AI 分析
项目摘要
Laya is an open-source, local-first AI notification command center that aggregates notifications from Slack, Gmail, GitHub, Jira, Notion, Outlook, and Calendar, using local LLMs (Ollama/LM Studio) or cloud models (BYOK) to pre-process them into ready-to-approve Action Cards. It features multi-persona AI routing, cross-platform entity resolution, and integrates with n8n for event normalization and action execution.
为什么值得关注
Laya combines local-first AI processing with broad professional tool integration to create a novel 'Action Card' paradigm where notifications are pre-researched and staged for approval before you even see them, potentially saving significant time for knowledge workers overwhelmed by multi-platform notifications.
优势
- Local-first architecture with Ollama/LM Studio support for privacy-conscious users
- Multi-persona AI routing that tailors processing to different professional roles (Engineer, Comms, Ops, etc.)
- Broad integration ecosystem covering Slack, Gmail, GitHub, Jira, Notion, Outlook, Calendar, and more
- Action Cards paradigm that pre-researches and stages notifications for one-click approval
- Cross-platform entity resolution linking related items across tools (e.g., Jira ticket to GitHub PR)
- Flexible model backends including local LLMs, cloud APIs, and CLI agent quotas via LiteLLM
局限性
- No automated tests, raising concerns about reliability and regression safety
- Single contributor creates bus factor risk and may limit development velocity
- No Docker support or official examples, increasing setup friction for new users
- Requires n8n as an external dependency, adding deployment complexity
使用场景
- Developers and engineers managing high-volume notifications across GitHub, Jira, and Slack
- Knowledge workers wanting AI-prioritized, pre-processed notifications from multiple platforms
- Privacy-conscious teams that need local LLM processing without sending data to cloud APIs
- Professionals seeking to reduce notification overload with automated research and action staging