neo4j-labs/llm-graph-builder — Neo4j graph construction from unstructured data using LLMs

Description and use

**Summary**
neo4j‑labs/llm‑graph‑builder is an open‑source toolkit that lets you turn unstructured text into Neo4j graph structures by orchestrating Large Language Model (LLM) agents, tool‑use pipelines, and memory management. It packages common prompting patterns into repeatable, multi‑agent workflows, making it easier to build knowledge graphs, RAG back‑ends, or data‑centric applications without hand‑crafting each LLM call.

**Value**
The project abstracts the boilerplate of LLM orchestration—prompt chaining, tool selection, and stateful memory—into reusable agents that directly emit Neo4j Cypher. This accelerates the creation of knowledge graphs from documents, emails, tickets, or any free‑form source, and provides a consistent interface for scaling from a single‑prompt prototype to a production‑grade pipeline.

**Practical adoption path**

1. **Proof‑of‑concept** – Clone the repo, run the provided Jupyter notebooks on a small dataset, and verify that the generated Cypher correctly populates a local Neo4j instance.
2. **Integration** – Wrap the notebook logic into a Python module or micro‑service, replace the demo LLM with your preferred provider (OpenAI, Anthropic, etc.), and expose a simple API for downstream apps.
3. **Scale & governance** – Add your own tool plugins (e.g., document loaders, entity extractors), configure persistent agent memory (Redis, Neo4j), and integrate with CI/CD for automated testing of graph updates.

**Production readiness**
The repository shows strong community signals (≈5 k stars, >800 forks, recent commits, and active issue discussion) and is written in Jupyter notebooks that are easy to translate into production code. While the integration steps are not fully documented, the codebase is modular enough to be containerized and deployed behind an API gateway. Starting with a small PoC and validating the setup cost will mitigate the main risk, after which the project is well‑positioned for a serious pilot in production environments.

Problem

neo4j-labs/llm-graph-builder helps turn isolated prompts and tools into repeatable agent workflows.

Outcome

High for an OSS candidate: recent activity, adoption, and ecosystem signals are strong enough for a serious pilot.

Workflow

• coordinate multi-agent workflows
• add tool-use pipelines
• standardize agent memory

Evidence

Neo4j graph construction from unstructured data using LLMs

Source