LangChain 4j runtime
Leverages the LangChain 4j library so you can compose complex agent chains natively in Java.
Bring the power of AI agents to business processes
The AI Agent connector is a LangChain 4j-powered service task extension for Camunda 8.8-alpha4. It allows you to embed goal-driven large language model (LLM) agents directly in BPMN while providing many benefits. Short-term conversation memory process variables are automatically injected into the agent's context and updated every turn. Guardrails and escalation sentiment, hallucination, and policy checks pause the flow and trigger a user task when thresholds are breached. Ad-hoc tool calls agents can request external actions (for example, Send Email, Write Memory, Query DB) that are fulfilled by child sub-processes created on demand. Please note that alpha software interfaces may change without notice.
Features and Benefits
Short-term memory support
Automatically captures process variables and re-injects them into every agent turn to preserve dialogue context.
BPMN-driven guardrails
Configure max-token consumption and escalation thresholds directly in BPMN; the agent pauses and creates a user task when limits are exceeded.
FEEL fromAi() function
Use the custom fromAi() FEEL function to obtain agent-generated text or structured JSON directly within BPMN expressions.
Ad-hoc subprocess tool calls
Agents can request any action modeled inside a BPMN ad-hoc sub-process, giving them access to all available tools without extra code.
Details
- Marketplace release date -
- Last Github commit -
-
Associated Product Group Categories:
- AI Services
-
Version Compatibility:
-
Used resources: