The Weekly Edge: Release of the Gremlins, Spotlight on Embeddable Graph Databases, & More [21 November 2025]

I’ve been in the media biz long enough to learn that sometimes you curate the news and sometimes the news curates you. This week is definitely the latter.

For most editions, the Weekly Edge is like a magazine of graph tech content filled with relevant-but-time-flexible reads, videos, and other resources curated by the team at G.V(). But this week, it’s a newspaper: All fresh headlines of breaking news and announcements. 

Here’s your front page of what’s happening in graph technology this week:

• Hello, world: GraphLite becomes the newest embeddable (+ GQL) graph database
• Grix & Gremfir: Apache TinkerPop pops out two new releases
• More graph, less database: PuppyGraph launches big integration with Microsoft OneLake
• Embeddable interviews: RyuGraph & LadybugDB leaders get in front of the camera
• Labeled Property Graphs 102: Deepening your understanding of this popular model 

Now, the news in full:

[News:] Meet GraphLite, the New(est) Embeddable Graph Database

In case you missed the memo, embeddable graph databases are all the rage right now, and this week, a new one just joined the club. GraphLite, is an open source embedded graph database implementing ISO Graph Query Language (GQL) in Rust. 

GraphLite was first developed by DeepGraph AI until they decided to open source it and release it to the community, according to the official announcement. Among all the embeddable graph databases seeking to fill the power vacuum left by Kuzu, GraphLite is the first one to implement GQL. 

Gajanan C., CTO of DeepGraph AI, shared that this choice to use GQL was intentional: “Just as SQL transformed relational databases, ISO GQL will do the same for graphs.” 

Welcome to the graph community, GraphLite!

[Release:] Apache TinkerPop 3.8.0 Introduces Gremlin MCP Server & More

If, like me, you’re new to the world of Apache TinkerPop™, then you’re in for a treat: Apparently each new version number gets a cute Gremlin nickname. 😍 This week saw the debut of the newest Gremlin to join the TinkerPop family: Grix Greven, a.k.a. TinkerPop 3.8.0.

There’s lots to the new release of this graph computing framework, most notably the introduction of a Gremlin MCP server and a turnip truckload of other new features and bug fixes. According to the release announcement, the main theme of 3.8.0 was improving the consistency and usability of the Gremlin query language. 

The Grix Greven release includes semantic changes to Gremlin – including repeat(), local(), and choose() – as well as new additions to the language including asNumber() and asBool() steps and the new P.typeOf() predicate. Many of these changes are breaking for certain usages, so users should consult the upgrade documentation carefully when upgrading to 3.8.0.

Long-time contributor Stephen Mallette shared some TinkerPop 3.8.0 reflections on his regular Gremlin Snippets blog, and the team at YouTrackDB have a great write-up on the top 10 new Gremlin 3.8 features and changes. But for getting into the nitty-gritty details, I found the best breakdown of the new release was written by newcomer contributor Andrea Child.

Also this week was the concurrent release of Apache TinkerPop 3.7.5, a.k.a. Gremfir Master of the Pan Flute 🤩. This was a maintenance release primarily consisting of bug fixes and small non-breaking improvements. 

[Launch:] PuppyGraph + Microsoft OneLake for Graph Analytics

At Microsoft Ignite earlier this week, Amir Netz, CTO of Microsoft Fabric, announced a raft full of new Microsoft OneLake integrations, including one for PuppyGraph. The new integration is powered by the OneLake Table API for Apache Iceberg and allows users to run real-time graph analytics directly on their Fabric Delta and/or Iceberg tables with zero ETL and no duplicate data (or duplicate database).

This short video gives you the tl;dr of the announcement, and this long-form blog post dives into the details with example queries and more. As a graph query engine – and not a graph database – PuppyGraph fills a unique niche within the graph tech ecosystem, so this new integration is another reason to keep watching this space.

[Watch:] Graph Chats with RyuGraph & Ladybug DB Founders

At ODSC earlier this month, Amy Hodler from GraphGeeks and I had the joy of sitting down (and/or standing up) with the leaders of two different embeddable graph database projects – both forks of Kuzu DB – as part of a Graph Chat video series sponsored by G.V(). 

Amy interviewed Akon Dey, Co-Founder of Predictable Labs, about his work with RyuGraph and its focus on knowledge and context management for AI and agents.