Agent Frameworks

crewAI v1.14.7a4 adds pluggable backends for memory, knowledge, and RAG

53.2k-starred framework now lets you swap backends for memory, knowledge, and flow orchestration.

Deep Dive

crewAI, the open-source multi-agent orchestration framework with 53.2k GitHub stars, has released pre-release version 1.14.7a4. This update introduces pluggable default backends for four key components: memory, knowledge, retrieval-augmented generation (RAG), and flow. Developers can now swap out these backends via configuration, making the framework more adaptable for custom storage or retrieval solutions.

Additionally, the @listen and @router runtime has been migrated to read from FlowDefinition, streamlining flow control within multi-agent systems. The pre-release also includes changelog and version updates from v1.14.7a3. Contributed by greysonlalonde, mattatcha, and vinibrsl, this iteration reinforces crewAI's position as a leading tool for building AI agent ecosystems.

Key Points
  • Pluggable default backends for memory, knowledge, RAG, and flow allow custom storage solutions.
  • Migrated @listen/@router runtime to read from FlowDefinition for better flow orchestration.
  • Pre-release v1.14.7a4 follows v1.14.7a3, with contributions from 3 developers.

Why It Matters

Makes crewAI more modular and enterprise-ready, letting teams integrate custom data pipelines and orchestration logic.

📬 Get the top 10 AI stories daily