Introducing Updated LangChain Integration and Memgraph Toolkit for Agents

In early 2023, LangChain made significant strides in providing abstractions for LLMs and query pipelines. As LangChain evolved, we introduced support for the MemgraphQA chain and Knowledge Graph construction.

With LangChain transitioning to LangGraph as part of its ongoing development, we have revisited our integration within the framework and made several updates.

Migration of the Integration

The Memgraph integration has been moved from the LangChain repository to a dedicated repository: langchain-memgraph. This change allows us to iterate more efficiently and deliver new features faster to both customers and the community using LangChain or LangGraph for agentic applications.

Additionally, the integration is now available on PyPI. To install it in your environment, simply run:

pip install langchain-memgraph

Once installed, the package includes both the MemgraphQA chain and the Knowledge Graph construction pipeline, as demonstrated in this example from our documentation.

Introducing the Memgraph Toolkit for Agents

Agentic applications rely on an LLM-based agent to determine the necessary actions to solve a given problem. To execute these actions, the agent requires a set of tools. This approach is fundamental to building agentic applications with LangGraph.

Memgraph now provides its own toolkit, fully compatible with LangGraph. The toolkit can support multiple tools, defined by either us or the community. Currently, we have added support for executing Cypher queries, with plans to expand the available tools over time.

Next Steps

In the near future, we plan to develop additional tools and integrate them into the Memgraph toolkit. As the toolkit evolves, we welcome feedback and suggestions on what to implement next. If you have ideas, feel free to reach out, contribute, or open an issue on the repository. We look forward to seeing what you build!