Kasal

Build intelligent AI agent workflows with visual simplicity and enterprise power.

Kasal turns complex AI orchestration into an intuitive visual experience. Design, deploy, and monitor autonomous AI agents that collaborate to solve real-world business problems — without writing orchestration code.

Why Kasal?

• Visual Workflow Designer — Drag-and-drop canvas for composing sophisticated agent interactions
• Enterprise-Ready — Built for Databricks with OAuth, teamspace isolation, and scale in mind
• Extensible Toolkit — A rich library of tools, including Genie, MCP servers, custom APIs, and data connectors
• Real-Time Monitoring — Live execution tracking with detailed logs, traces, and performance insights
• Production-Grade — Robust error handling, retry logic, and enterprise deployment patterns

What You Can Build

• Data Analysis Pipelines — Agents that query, analyze, and visualize your data
• Content Generation Systems — Collaborative agents for research, writing, and content creation
• Business Process Automation — Intelligent workflows that adapt and make decisions
• Customer Support Assistants — Multi-agent systems with specialized knowledge domains
• Research & Development — Agents that gather, synthesize, and present insights

Get Started in Minutes

Databricks Marketplace (Recommended)

Install directly from the Databricks Apps Marketplace with one click — the best path for production, with automatic updates and enterprise support.

Deploy from Source

Use the deployment script in this repository for custom installations. Ideal for tailored configurations and advanced setups.

Local Development

A quick setup for testing and development — requires Python 3.9+ and Node.js.

See It in Action

The visual workflow designer for building AI agent collaborations

Create your first agent workflow in under two minutes:

1. Design — Drag agents onto the canvas and define their roles
2. Connect — Link agents together to form collaboration flows
3. Execute — Hit run and watch your agents work as a team
4. Monitor — Follow real-time logs and execution traces

Documentation

More Documentation

• Docs Hub - Documentation index
• End‑User Tutorial Catalog - Screenshot-ready walkthroughs
• Testing Guide - Testing strategy and coverage

Architecture

Kasal follows a clean, layered architecture designed for scalability and maintainability:

Frontend (React) → API (FastAPI) → Services → Repositories → Database

The CrewAI engine plugs in at the service layer to drive intelligent agent orchestration.

Known Limitations

Entity Memory with Specific Models

Entity extraction in memory backends has known compatibility issues with:

• Databricks Claude (databricks-claude-*) — JSON schema validation errors
• Databricks GPT-OSS (databricks-gpt-oss-*) — empty response errors

Automatic fallback: When these models are detected, Kasal transparently uses databricks-llama-4-maverick for entity extraction while keeping your chosen model for every other agent task.

License

Licensed under the Databricks License

Additional Resources

Unlocking Databricks Marketplace: A Hands-On Guide for Data Consumers