Web portal users

The October 2026 release for web portal users adds a new chat page for working with NexusOne’s AI. It has search, file attachments, shared workspaces, and read-only share links. This release also fixes portal-generated and manually deployed Airflow DAGs from colliding with each other. It adds AWS-hosted Claude endpoints, AI-assisted Ranger policy suggestions, spreadsheet-authored business rules, Superset chart exports, S3-backed Elyra pipelines, and migration feature configuration changes.

​

New features

This section contains new features recently added to the NexusOne platform.

​

Chat

Chat is a new page in the portal where you talk to NexusOne’s AI directly. It lets you do the following:

• Search: Search your past conversations to find one again.
• File attachments: Paste or attach a text file, such as a spreadsheet, directly in a conversation.
• Shared workspaces: Share a chat workspace with your team, the same way you’d share a business rules workspace.
• Read-only share links: Generate a link to a conversation for others to view without letting them edit it.

​

Bug fixes

This section contains fixes for issues affecting apps or features on the NexusOne platform.

​

Portal-generated Airflow DAGs

Airflow DAGs on NexusOne come from two sources:

• Portal-generated: The NexusOne portal creates some DAGs for you automatically.
• Manually deployed: You write and deploy your own DAGs in either of these ways:
  • Upload DAG files to the Airflow DAGs folder in S3.
  • Publish a pipeline’s DAG from Elyra.

This release fixes two issues that could arise from mixing portal-generated and manually deployed DAGs. The following sections cover each one.

​

Overwriting manually deployed DAGs

Portal-generated and manually deployed DAGs used to share the same storage location, called a DAG bundle. Redeploying one with the same name could silently overwrite the other. In this release, portal-generated DAGs now live in their own bundle, separate from anything you deploy manually.

​

Sharing Airflow variables with manual runs

A portal or manually triggered run of the same DAG used to share the same Airflow variables. One could silently pick up the other’s settings, even mid-run. This release writes every variable the portal owns under an nx1_ prefix, scoped to the portal run. A manual run never reads a portal-owned variable.

​

Enhancements

This section contains enhancements to existing app features on the NexusOne platform.

​

AI-assisted policy generation

Sensitive columns, like ones tagged PII or Payment Card Industry (PCI), need a Ranger policy to mask or restrict them. Writing one by hand for every column doesn’t scale. NexusOne can now draft that policy for you automatically. However, nothing reaches Ranger until you review and accept a suggestion. You can reach the feature at Govern > Data access. Request suggestions for a table or a whole schema, then accept or reject them at the batch or individual level.

​

Business rules now support spreadsheet authoring

Business rules, introduced in nx1-deployer image v1.16.1, now lets you author a rule from a spreadsheet or a rule table. Before, you could only build one in the rule editor. You can also grant a Keycloak role access to a workspace, instead of adding members one at a time. Anyone who holds that role gets access to the workspace. This sits alongside the existing per-member workspace roles described in Business rules.

​

Connect Claude through your AWS account

The AI Control Plane’s Endpoints editor now lets you add an endpoint for Claude Platform on AWS or Amazon Bedrock, instead of only the public Anthropic API. This matters if your AI traffic needs to stay inside your own cloud for compliance reasons. You can reach it at AI > AI Control Plane > Admin > Add endpoint. It includes the following new presets:

• Claude Platform on AWS
• Claude in Amazon Bedrock
• Bedrock, OpenAI-compatible

See Use Claude through your AWS account for how an administrator sets the initial Claude on AWS endpoint at deploy time.

​

Crew exports now include Superset charts

Crew results exported as PDF, described in Export a crew run or result, now embed a crew’s Superset charts as images, instead of just linking to them.

​

Elyra pipelines now publish to S3

The Elyra pipeline editor in JupyterHub now publishes a pipeline’s generated Airflow DAG to S3, instead of pushing it to a git repository. NexusOne resolves object storage credentials per bucket automatically, so you no longer configure them in the pipeline editor yourself.

​

Migration feature

This release adds configuration changes to how you set up and trigger a migration through the portal. Migration describes behavior changes to the migration DAGs themselves.

• Multi-tenant cluster profiles: The Infrastructure Config page’s Tenant Profiles table can now hold more than one source cluster profile, one row per tenant name. Pick which one to use for a run, right from the portal wizard.
• Service account identity: The Infrastructure Config page now has a Service Account User ID field, replacing the older MapR User field. It’s the username a migration logs in as over SSH on the source cluster, so the migration only gets that user’s access.
• Configurable cluster paths: The Infrastructure Config page’s Cluster & Edge Node Paths section adds three new fields:
  • Cluster Edge Discovery Temp Path
  • Cluster Hive Scratch Dir
  • Cluster DistCp Log Root
  Each is now set per tenant, instead of sharing one value across every tenant.