Invitation] Join OpenCrawling: Help Us Build Next-Gen Process-Aware AI & RAG for Flowable!

Flowable Engine
1

Hi Flowable Community! :waving_hand:

We are huge fans of the Flowable open source ecosystem and the incredible work done here around BPMN, CMMN, and DMN process automation.

Today, we would like to invite developers, architects, and contributors from the Flowable community to collaborate on OpenCrawling an open source, event-driven data ingestion and crawling framework built for Enterprise RAG (Retrieval-Augmented Generation), Vector Search, and Model Context Protocol (MCP) AI Agents.

With the release of our dedicated module, oc-flowable-repository-connector, OpenCrawling now natively connects to Flowable engines to bridge the gap between business process execution and enterprise Large Language Models (LLMs).

:light_bulb: WHY PROCESS-AWARE AI NEEDS FLOWABLE EXPERTISE

While modern AI assistants excel at searching static documentation or PDF policies, they often lack visibility into live enterprise process executions, historic task variables, and candidate group security.

OpenCrawling extracts BPMN 2.0 process definitions, historic process instance variables, user task forms, and candidate group permissions into standardized Open Ingestion Standard (OIS) document streams.

This enables AI agents to accurately answer questions like:
• “What is the current bottleneck in the Loan Approval workflow?”
• “Show me all process executions associated with Client #9402.”
• “Which candidate groups are authorized to inspect these active tasks?”

:rocket: HOW THE CONNECTOR WORKS (CURRENT ARCHITECTURE)

• Target Engine: Flowable REST API (Process, CMMN, DMN)
• Ingestion Pipeline: BPMN XML models → OIS Document Schemas → Apache Kafka → Chunking & Ollama/Vector Store Embeddings (pgvector, Qdrant, Elasticsearch)
• Security: Maps Flowable Candidate Groups (e.g., group:finance-approvers) to document Access Control Lists (ACLs) for zero-trust RAG filtering.
• Core Tech Stack: Java 25, Virtual Threads, Structured Task Scope, Spring Boot 3.4+, Apache Kafka.

:handshake: CONTRIBUTION OPPORTUNITIES: WHERE WE NEED YOUR FLOWABLE EXPERTISE!

We are actively seeking contributors from the Flowable community to help push the boundaries of process-aware AI. Whether you are a core BPMN specialist, Java developer, or enterprise architect, there are exciting areas to build together:

  1. Deep Engine Integrations & Event Listeners:

    • Building native Flowable Event Listener extensions for real-time process state change streaming (bypassing REST polling).
    • Expanding extraction for complex CMMN case models and DMN decision table executions.

  2. Variable Transformation & Prompt Engineering:

    • Enhancing BPMN 2.0 XML parsing into natural language narratives tailored for LLM context windows.
    • Intelligent serialization of complex serializable process variables and JSON form definitions.

  3. Advanced Identity & ACL Mapping:

    • Integrating Flowable IdmEngine, Keycloak, and LDAP identity resolution into OpenCrawling’s unified security context.

  4. Performance & Scalability:

    • Optimizing high-throughput historical data extractions on multi-million instance Flowable databases using Java 25 concurrency models.

:link: RESOURCES & GETTING STARTED

• Blog Announcement: Natively Connecting Flowable & Camunda BPMN Engines to Enterprise RAG | OpenCrawling
• OpenCrawling GitHub: Open Crawling · GitHub

If you are interested in contributing, feel free to drop a message in this thread, open a GitHub Issue/PR, or star the project on GitHub!

Let’s build the future of Process-Aware Enterprise AI together! :rocket: