Features
Visual drag-and-drop workflow editor (Vue Flow)
AI assistant generates workflows from natural language or voice
Multi-agent orchestration with up to 5 levels of nesting
Built-in RAG pipeline (Qdrant or pgvector) with document upload
MCP (Model Context Protocol) client and server support
Human-in-the-loop approval checkpoints
Content guardrails with 4 layers (input, output, action, human review)
Persistent agent memory with knowledge graphs
Execution traces and evals for debugging and testing
Token cost tracking per workflow
User-built dashboards with chart widgets (Dashboard tab)
Python tools and skills system for portable agent capabilities
Playwright browser automation with auto-healing
Data pinning and version history for workflows
Parallel DAG execution for independent branches
Packet-Aware AI Agents for cross-document validation
Multi-Model Optimization for dynamic model selection
Deep Document Understanding for structured document maps
Contextual linking across documents in a packet
Applied business logic for rule-based analysis
Auditable and verifiable generative AI outputs
Human-in-the-loop review with task dashboards
No-code app builder for document processing workflows
Chat with documents for on-demand insights (single & multi-file)
Supports 55+ rich file formats
Handwriting recognition and multi-language recognition
Long document support (100+ pages)
Unlimited users in collaborative and personal workspaces
Enterprise-grade security with compliance controls (SSO, IP whitelisting)
Connect your own cloud storage (AWS S3, Azure, GCP, Google Drive)
