On November 9, 2026, AWS published an open-source tool with a local browser interface that moves selected data lake records into Amazon Neptune Analytics as a graph. Users point it at an Athena catalog, describe the graph they want in SQL/PGQ, check a sample of the rows and then load the result, without writing an ETL pipeline by hand.
The tool matters for data teams that keep large Apache Iceberg tables in S3 but want to ask relationship questions, such as which accounts are linked through shared transactions. Until now that kind of analysis usually meant several manual steps across Athena, S3 and Neptune. The new UI wraps those steps in one guided flow.
What AWS Announced
The release is a local web application, not a new managed service. AWS describes it as a way to project chosen tables and relationships from a data lake into a graph, then run graph algorithms on that projection in Neptune Analytics.
AWS gives a fraud-style example: load accounts and the transactions between them, then look for clusters of connected accounts that may deserve a closer look. The point is that you only load the slice you need for one question, rather than copying a whole dataset into the graph engine.
The code lives in the awslabs nx-neptune repository on GitHub, in a folder called proxy. AWS says the UI works in every region where Neptune Analytics is offered.
How the Workflow Works
The flow follows a fixed order of steps. Based on the announcement, it looks like this:
- Pick an Athena catalog and database that hold your Iceberg tables.
- Define a property graph with SQL/PGQ, mapping tables to nodes and edges.
- Preview sample rows before anything is imported.
- Let the tool run the projection queries in Athena and write the results to an S3 staging area.
- Import the staged files into a Neptune Analytics graph that the UI then helps you manage.
- Explore the finished graph visually and run queries against it with Graph Explorer.
Because the projection runs in Athena, the heavy filtering happens before data reaches Neptune. That keeps the graph small and focused, which also helps control cost.
Under the Hood
The GitHub README shows a simple stack: a Python FastAPI backend that drives Athena queries and the life cycle of Neptune graphs. The frontend is a React and TypeScript single-page app built with Vite. Project and projection details are kept in a local SQLite file under the user's home folder.
Setup needs Python 3.11 or newer, Node.js with npm for the frontend build, plus AWS credentials allowed to use Athena, S3 and Neptune Analytics. Make targets install the package, compile the frontend and start the server on port 8080 bound to localhost. A Docker Compose option starts the server together with Graph Explorer in containers.
Each server start prints a launch link with a one-time access token, and API calls fail without it. The README also warns that the quick development targets create graphs that are reachable publicly and lack deletion protection, so they are meant only for test data. Running the server module directly without those settings keeps the safer defaults.
What It Means for Developers
For database developers, the most interesting part is SQL/PGQ. It is the property graph part of the SQL standard, and Oracle Database also supports it. Teams that already model graphs over relational tables will find the same idea here, applied to Iceberg tables through Athena.
The tool also lowers the bar for trying graph analytics. You can test whether a graph view of your data reveals anything useful before building a production pipeline. If you already run an open lakehouse, our story on Oracle AI Lakehouse and Iceberg catalog federation shows how other vendors are approaching the same tables.
Treat the security notes seriously. Use short-lived credentials from SSO or an assumed role, never long-lived IAM user keys. Keep the server on localhost, and do not reuse the Compose setup on shared or production hosts.
How to Get Started
Clone the awslabs nx-neptune repository and follow the setup guide linked from the proxy README. Install the library from source, build the UI, then start the development server and open the launch link it prints. Begin with a small, non-sensitive Iceberg table so you can see the full path from Athena to Graph Explorer.
Check your IAM policy before the first run, because the tool needs permissions in three services. Also plan to delete test graphs when you finish, since Neptune Analytics charges while a graph exists.
Conclusion
The new open-source UI gives AWS users a guided route from Iceberg tables to Neptune Analytics graphs, using SQL/PGQ for the mapping and Athena for the heavy lifting. It is a practical way to test graph questions on lake data, as long as you keep it on a developer machine with short-lived credentials.
Sources
- AWS What's New: Introducing an open-source UI for graph analytics over data lakes with Amazon Neptune Analytics
- awslabs/nx-neptune on GitHub: proxy README
See my articles more often when you search Google.
