[Feature Request] Introduce `simple-graph-rag` Sample Application as Stepping Stone to Agentic GraphRAG

[ishaanv1709]

Hi @bharagha @14pankaj @jgespino, I have a proposal for following feature, please assign this issue to me

Feature Description

Abstract

I propose creating a new simple-graph-rag sample application as a foundational step prior to fully developing the complex "Agentic GraphRAG" project (LangGraph/Supervisor architecture). This application will serve to correctly implement, isolate, and test the base-level integration between the OpenVINO hardware inference loop and Neo4j Text-to-Cypher generation on Edge hardware.

Why this stepping stone is critical:

The ultimate goal of the GSoC project is a fully autonomous Agentic GraphRAG system utilizing LangGraph and a Supervisor routing structure. However, jumping directly from the existing baseline chat-question-and-answer (which relies purely on vector-based semantic search) to a complex multi-agent orchestration generating dynamic Cypher queries represents a massive architectural leap.

Currently, the open-edge ecosystem lacks a standardized, foundational mechanism to reliably ping and query Neo4j databases from an OpenVINO inference loop.

Therefore, building a pure baseline simple-graph-rag sample application alongside the existing directory is essential. By deliberately abstracting away the LangGraph routing overhead, this application will:

1. Isolate and test the raw OpenVINO to Neo4j Text-to-Cypher generation flow on edge hardware.
2. Provide a clean, proven baseline connection that can be merged and verified.
3. Set a much safer stage to build the complex Agentic supervisor topology upon later, without forcing developers to troubleshoot the entire AI stack at once.

Proposed Solution

1. Create a new directory sample-applications/simple-graph-rag/.
2. Implement a standard (non-agentic) Linear Chain that translates a user query explicitly into Cypher utilizing an LLM, executes the query on a local containerized Neo4j database, and synthesizes the context.
3. Keep the flow purely focused on Graph interaction, providing a smooth transition path from the current vector-only implementations to the final Agentic vision.