Agents Icon

AI Agents & Workflows

  • Individual Agents:Custom agents with configurable system messages, tool selection, model parameters, and conversation styles
  • Conversation Styles:Concise, Detailed, Casual, Professional, or Custom
  • Operational Modes:
    • - Plan Mode:Read-only structured planning
    • - Verify Mode:Systematic requirement validation
    • - Default Mode:Full tool access with natural agent behavior
Chat Icon

Chat & Sessions

  • Real-time chat with streaming responses
  • Session branching:Fork a conversation at any message to explore alternatives
  • Pause / Resume:Suspend and continue agent execution
  • Session copying with configuration updates
  • Parent-child branch tracking
  • Message retrieval, deletion, and cross-session persistence
  • Auto-cleanup of old sessions
Knowledge Base Icon

Knowledge Base

  • Upload documents to a shared team knowledge base
  • Semantic search with natural language queries
  • Document codes for referencing and filtering
  • Data access groups for permission control
  • Agentic PLAN-ACT-REASON processing for complex queries
  • Result caching with freshness validation
  • Citation tracking by document code/name
  • Visual analysis of charts and diagrams in PDFs
  • Supported formats: Markdown, PDF, Word, Excel, Images, Code
Skills System Icon

Skills System

  • Reusable code and prompt library shared across agents and teams
  • Skill types: script, prompt, other
  • Automatically available to agents when assigned
  • Data access groups for permission control
  • Modification tracking (creator, modifier, timestamps)
  • Automatic sync to workflows and agents
Model Management Icon

Model Management

20+ models across major providers with intelligent routing. Features include per-user model override, privacy-based defaults, cost/quality/speed scoring, custom model registration, and extended thinking support for reasoning models.

Anthropic Claude

  • Claude Opus 4.6:flagship reasoning
  • Claude Sonnet 4.6:balanced production
  • Claude Haiku 4.5:fast and low-cost

OpenAI GPT

  • GPT-5.4:deep reasoning
  • GPT-5.3 Codex:frontier coding
  • GPT-5 Nano:ultra-low-latency
  • GPT-4o Search Preview:web search optimized
  • GPT Image 1.5:text rendering in images

Google Gemini

  • Gemini 3.1 Pro Preview:multimodal reasoning
  • Gemini 3 Flash Preview:fast and cost-effective
  • Gemini 2.5 Computer Use:automation

Specialized

  • Auto:intelligent model selection
  • Auto Fast:ultra-low-latency routing
  • GPT-OSS 120B:open-source 120B model
  • Veo 3.1:professional video generation
Built-in Tools Icon

Built-in Tools

Research & Analysis

  • Web search with source citations
  • Fetch and parse web pages
  • Multi-step deep research with analysis
  • Semantic search across uploaded documents
  • Extract content from DOCX, XLSX, PDF, images
  • AI-powered file analysis

Code Execution

  • Run Python, Bash, JavaScript, PowerShell, HTML/CSS
  • Read, write, and edit workspace files
  • Syntax-highlighted code display
  • Shell commands and running service detection

Media Generation

  • Text-to-image with reference image support
  • Image editing, variations, and mask-based edits
  • Image format conversion (HEIC, WebP, RAW)
  • Video generation:up to 60s at 1080p with audio
  • Screen capture

Computer Use

  • Vision-based desktop control
  • Browser automation across multiple providers
  • Sandboxed computer control
  • Chrome extension for direct browser control
  • Multi-tab support and visual feedback loops

Orchestration

  • Delegate to specialized or named agents
  • Distribute work across teams
  • Persistent agent memory
  • Reusable skill library
  • Task management and progress tracking
  • Email, scheduling, and user input prompts
Deployment Options Icon

Deployment Options

Desktop Application

  • macOS:Apple Silicon and Intel
  • Windows:Native
  • Automatic updates
  • Multi-environment support

Web / Cloud

  • Full REST API
  • Real-time streaming chat
  • Secure authentication
  • Intelligent model routing

Sandboxed Execution

  • Isolated code execution environment
  • Secure sandboxing for untrusted code
  • Automatic resource management
Security & Collaboration Icon

Security & Collaboration

  • Authentication: Secure login with token and API key support
  • Authorization: Role-based access control
  • Data access groups: Fine-grained document and skill visibility
  • Code execution safety: AI-powered risk assessment before running code
  • Audit trail: Request/action logging with user tracking
  • Team management: Users, teams, agents, workflows
Scheduling & Billing Icon

Scheduling & Billing

  • Scheduled agent executions on recurring or one-time schedules
  • Automated task processing
  • Credit-based pricing with secure payment processing
  • Usage tracking per user and team
  • Cost transparency per model