Linkurious Enterprise 4.3: Use natural language questions to explore your graph, with Query AI

The release of Linkurious Enterprise 4.3 focuses on making investigations faster and more comfortable, with a stable Query AI experience, major Property Panel usability improvements, and CSV Import — alongside a few important platform updates to be aware of.

Query AI is now officially stable and ready for everyday investigative workflows.

It lets you explore your graph using natural language questions, instantly generating Cypher queries that you can review, edit, and save, keeping you fully in control.

Whether you’re an investigator, analyst, or occasional graph user, Query AI helps you reach insights faster without requiring deep Cypher expertise. It works with your preferred LLM (ChatGPT, Gemini, Claude, and more), making graph exploration more intuitive and accessible.

This release brings a set of thoughtful improvements to the Property Panel, designed to reduce friction and help you stay focused on what matters most during investigations.

What can you expect from these Property Panel improvements?

The Property Panel is now collapsed by default when selecting nodes or edges.

This gives you more space to focus on the graph itself, especially in dense or highly connected datasets. When you need details, the panel is always one click away—and it stays exactly how you left it while working within the same visualization.

Working with multiple nodes or edges is now clearer and more intuitive.

You can easily locate selected entities directly from the Property Panel and visually identify them in the graph, helping you maintain context when analyzing large tables or complex selections.

Reviewing data is now more comfortable thanks to several usability improvements:

• Long property values are easier to read
• Table headers remain visible while scrolling
• Navigating between property categories is faster and smoother

The Property Panel can now be resized, letting you adapt the interface to your workflow, whether you want to prioritize detailed data inspection or keep the graph front and center.

Bringing external data into your investigations is easier than ever thanks to the new CSV Import feature. Because CSV is a universal format supported by nearly every system, users can quickly upload and explore new datasets without any technical setup or connectors. This enables easy integration with external tools that don’t natively connect to Linkurious Enterprise.

CSV import also unlocks rapid prototyping: investigators can test hypotheses, visualize relationships, or enrich an existing graph in minutes, without waiting for a full data source to be configured. For example, during a fraud investigation involving hundreds of customers, analysts can import phone call records received as CSV files and link them to existing phone number nodes, instantly revealing hidden connections and new investigative leads.

As part of this minor release, a few changes may require attention depending on your setup:

• Browser support has been updated. Versions released more than 2 years ago are no longer supported. Please ensure you are using Chrome 119+, Edge 119+, or Firefox 120+ for the best experience.
• The minimum supported Neo4j version is now 4.4.0
• Elasticsearch 7 is no longer supported. Deployments must use Elasticsearch 8+.
• Embedded Elasticsearch is no longer available on macOS. If you run Linkurious Enterprise on macOS, you’ll need to host Elasticsearch separately. Please refer directly to the ElasticSearch deployment documentation for help.
• The Image Export plugin is no longer maintained. You can still use it through custom actions, but we recommend using the native export feature instead.
• For new deployments, local account passwords must now have at least 12 characters
• Session timeout behavior updated. The access.loginTimeout setting now defaults to 3 hours. If this parameter is not set, all user sessions will expire after 3 hours of inactivity.

These changes help ensure better security, long-term maintainability, and platform consistency.

Originally published 27 January 2026. Last updated 23 February 2026.