Features
Raw radar output as imagery and point clouds
Generative simulator for synthetic raw radar data
Integration with digital twin world models
Scene augmentation: add objects, change weather, modify behavior
Closed-loop evaluation for driving policies
Real-world data collection via vehicle-mounted radar
Edge case generation for adversarial scenarios
Designed for end-to-end neural network training
AI agents with email, SMS, and chat interfaces
Real browser automation with login support
Persistent built-in database per agent
Deliverables: reports, charts, PDFs, CSV exports
Pre-built workers for recruiting, sales, ops, risk, research
Custom instructions at personal and org level
Agent identity (name, email, phone number)
Per-agent sandboxing and encrypted secrets
Full audit trails and approval loops
Self-hosted deployment (MIT license)
Batch scheduling and queueing
Event triggers for automated workflows
Real browser sessions: click, type, navigate
File attachments: send and receive files
