The Weekly Edge: Graph Tech in Flux, Flexible GraphRAG, KGs for Synthetic Data & More [10 October 2025]

It’s 10/10, so the Weekly Edge is back with another 10 out of 10 edition.

This weekly blog series is like your fyp of videos, reads, repos, and more from across the graph technology ecosystem in the past seven-ish days, curated by the team at gdotv. 

Here’s this week’s headlines at a glance:

• New talks and slides have been uploaded from last month’s Graph Data Council meeting
• A new paper demos using a Neo4j knowledge graph to help generate synthetic data 
• A valuable repo for using GraphRAG across multiple vector and graph databases
• More details are now available about Graphs in Microsoft Fabric

Now, let’s dive into the deets.

[Short Watch:] GDC Talk: Graph Technology – A World in Flux

A month ago, the Graph Data Council (the council formerly known as the Linked Data Benchmark Council), held its 20th Technical User Community meeting, which hosted a turnip-truck full of graph talks across a wide range of industry topics. The news? Most of the videos and slides have now been published!

Out of all the presentations, I highly recommend the one from Max Latey, CEO & Founder of Pinboard Consulting, in which he gives a satellite’s-eye view of the graph tech ecosystem – where it’s been and where it’s going. It’s an optimistic take on everything from NetworkX and the Graph Query Language (GQL) standard to industry growth and new graph databases entering the market without glossing over where the industry needs to do better (outreach and growth) or find new opportunities to succeed (like graph on relational or JSON data stores). Catch the video above and grab your copy of the slides here.

Gdotv Supreme Leader & Founder Arthur Bigeard also gave a talk on a graph-native IDE for GQL (+ slides), demonstrating for you what it’s like to use G.V() with the industry-standard query language (spoilers: it’s awesome).

[Deep Read:] Generating Authentic Grounded Synthetic Maintenance Work Orders (Using Graphs)

If that title didn’t absolutely arrest your attention, then let me remind you: The reason tech usually solves the flashy problems before the boring problems is that the boring problems are often the hardest – and yet the most important – to solve. This new paper published in IEEE Xplore looks at the boring-but-important challenge of generating synthetic technical data for industrial maintenance, where real datasets are often limited and unbalanced.

Students and researchers at the University of Western Australia used a Neo4j knowledge graph and an industry-trained LLM to generate technical maintenance work orders that were nearly impossible to distinguish from their human-written counterparts. Read a tl;dr of the paper’s key findings from one of the researchers or dive into the complete code and datasets on GitHub.

And thanks to Paco Nathan for sharing!

[Repo:] Flexible GraphRAG with Any Graph Database

GraphRAG is all the rage these days, so naturally when we stumbled across this GitHub repo, it was a must-share. Flexible GraphRAG is a set of tools that support document processing, automated knowledge graph building, (Graph) RAG setup, and hybrid search that combines full-text, vector, and graph technologies, and the cherry on top: AI-powered Q&A chat. Like we said, it was a must-share.

Flexible GraphRAG is maintained by Steve Reiner from Integrated Semantics and allows you to work with multiple different vector, search, and graph databases as well as a number of LLMs, APIs, and more. Supported graph database include:

• Neo4j 
• Kuzu
• FalkorDB 
• ArcadeDB
• Memgraph 
• NebulaGraph 
• Amazon Neptune (& Neptune Analytics)

Like the name says: flexible. 

[Catch-Up/ICYMI:] Updates on Microsoft Graph in Fabric

One month ago in the Weekly Edge, I shared the news about the new Graph in Fabric capabilities announced by the Microsoft Fabric team at FabCon Europe. It had all the trappings of a conference keynote announcement, including a summary slide, pretty pictures, and not much substance. Now the details are starting to roll in. 

My former colleague Justin Fine shared a great tl;dr of updates on Graphs in Fabric on LinkedIn, which included this lengthy feature breakdown blog post and a quick-bite demo video. If you were wondering what all the hubbub might have been about and how you can get your hands on it, then these are resources worth checking out. 

P.S. There’s a new version of G.V() out in the wild!

That’s it for this week’s edition. Got something you want to nominate for inclusion in a future edition of the Weekly Edge? Ping us on on X | Bluesky | LinkedIn or email weeklyedge@gdotv.com.