create-agent-graph

Build multi-agent workflows with conditional routing in Kotlin using the ConditionalAgentGraphBuilder class, which enables dynamic branching and decision-making. It supports programmatic graph construction for deterministic execution across complex AI pipelines.

The best way to orchestrate sequential multi-agent pipelines is using the SimpleAgentGraphBuilder class to construct linear execution graphs. This approach handles deterministic data processing pipelines, content creation workflows, and multi-stage approval processes.

Yes, you can implement fan-out and fan-in patterns for dynamic decision-making using the ConditionalAgentGraphBuilder class. It provides specific code patterns for conditional branching, validation, error handling, and retries within complex agent workflows.

Yes, this multi-agent orchestration approach supports customer support routing workflows through the ConditionalAgentGraphBuilder class. It enables dynamic branching to route tasks across specialized agents based on conditional logic and validation.

A key limitation is that agent graphs require manual orchestration design for complex sequential and conditional steps. While builder APIs provide error handling and retry patterns, graph execution remains deterministic, limiting highly unpredictable dynamic routing scenarios.

Use KAgentic for multi-agent workflow orchestration to simplify the difficult manual orchestration of sequential and conditional steps. It provides clear guidance, builder APIs, and best practice patterns for error handling, retries, and validation in agent graphs.