| Key features | - 16 distinct reasoning strategies (ReAct, Chain-of-Thought, Tree-of-Thoughts, Reflexion, Plan & Execute, multi-agent, and more)
- Human-in-the-loop approval wired into live execution path
- Risk-classified tool policies (low, medium, high, critical)
- Real multi-tenant isolation with row-level security
- Full audit trail of logins, executions, tool calls, approvals, and configuration changes
- Self-extending tools: agents can propose new tool definitions at runtime
- Compliance evidence reporting for HIPAA, SOC2, and GDPR
| - Works in its own browser without accessing your personal browser or accounts
- Voice input with on-device speech recognition
- Projects to organize related tasks and share files between them
- Specialists: pre-configured versions for specific types of work like research or bookkeeping
- Routines that run automatically on a schedule (daily, weekly, or monthly)
- Integrations with Outlook, Teams, Numbers, Pages, Notes, Calendar, Reminders, Notion, Todoist, Linear, GitHub, and Stripe
- Local memory storage with Qdrant that is editable and visible
- Choice of local AI models (Qwen, Granite, gpt-oss) or Ollama, LM Studio, or online providers with your API key
- Chrome extension for optional access to your Chrome browser on approved sites
|
|---|
| What makes it different | - Agents reason through 16 strategies instead of following fixed scripts
- Human approval wired into execution path, not a separate setting or dashboard
- One integrated platform (builder, reasoning, governance, audit) instead of stitching together multiple vendors
- Open-source tool engine (Matimo OSS) so no vendor lock-in
- Bring your own LLM keys; no markup on model costs
| - Runs entirely locally on your Mac with no cloud storage or account required, keeping all data private by default
- Costs significantly less than cloud alternatives at $120-$300 per month
- Voice processing happens on device with audio never sent or stored
- Fully customizable with specialists you create for your own workflows
|