Latest updates from the Nexadata team
Product update

🌿 Mid-Spring Release: Connect More, Control More

Quin avatar
Shared by Quin • April 16, 2026

Hi there,

This release expands Nexadata's reach into enterprise environments. Connector Copilot now supports OData APIs natively, introduces a full Zero Data Retention (ZDR) compliance infrastructure, and adds a new transformation that gives you precise control over record identity in downstream systems. Here's what's new in the Mid-Spring Release.


🔌 OData Support in Connector Copilot

Connector Copilot now natively supports OData APIs, complementing the existing OpenAPI spec-based approach. Systems like Microsoft Dynamics, SAP, and Acumatica expose their data through OData, and Connector Copilot now supports OData versions V2, V3, and V4.

True to its agentic nature, the Connector Copilot autonomously parses the OData metadata, determines how to traverse the service, constructs valid requests, and builds a fully configured dataset. No manual spec work, no custom code.

How It Works

Once you provide an OData service URL, the Connector Copilot will:

  • Parse the OData $metadata document to discover entity types and properties
  • Intelligently handle OData-style pagination ($top, $skip, nextLink)
  • Construct valid OData requests to return structured, analytics-ready data
  • Apply the same natural language dataset configuration experience you already know

🔐 Full Zero Data Retention (ZDR) Compliance

Enterprise customers with strict data residency and retention requirements now have a clear, auditable path to deploy Nexadata in a fully ZDR-compliant configuration. This release delivers ZDR support across two layers: at the platform level and at the AI infrastructure level.

Platform-Level ZDR: Prompt Non-Persistence

Nexadata now supports a configurable ZDR mode that prevents end-user prompt content from being stored at rest. When ZDR mode is enabled, prompts submitted through the Copilot Suite are used in-flight for AI reasoning and discarded. They are never written to persistent storage.

This directly addresses a compliance barrier that enterprise customers face when deploying AI-assisted tools in regulated environments.

View ZDR documentation →

Infrastructure-Level ZDR: Amazon Bedrock Support

Nexadata now supports Amazon Bedrock as an alternative AI inference provider. This enables per-request ZDR without requiring a separate enterprise contract, and gives compliance teams an additional lever to meet their data governance requirements.

Nexadata's core architecture is unchanged: Claude continues to serve as the reasoning layer, and all data processing remains within Nexadata's deterministic execution layer.


🆔 Generate Unique ID Transformation

A new transformation is now available: Generate Unique ID. This complements the existing Add Row Numbers transformation by generating a globally unique identifier for each record rather than a sequential row number.

Unlike row IDs, which are sequential and scoped to a single run, unique IDs are:

  • Collision-resistant across runs, environments, and source systems
  • Suitable as primary keys in downstream databases and data warehouses
  • Useful for deduplication when records from multiple sources need a shared reference key

View Generate Unique ID documentation →


🧱 Platform Fixes

  • ENG-910 / Bug Fix: Workflows no longer hang when a Primary pipeline is reassigned to a Supporting role. The platform now validates the pipeline designation before execution and surfaces an error before any hang can occur.

🔭 What's Next

We're continuing to push the boundaries of what Connector Copilot can connect to, and investing in more powerful pipeline authoring. Here's what we're working on:

  • Build Integrations Directly from API Documentation: Right now, Connector Copilot requires a formal API spec to build a connector. Many systems don't provide one. In the next release, Connector Copilot will be able to read a system's documentation website -- the same pages a developer would read -- and autonomously extract the knowledge needed to build a working connector. Think of it like the scene in The Matrix where Neo downloads Kung Fu: rather than needing a structured feed, the agent comprehends unstructured information and emerges ready to act. This capability has the potential to open Connector Copilot to a near-endless range of integrations.
  • Holistic Pipeline Editing: Refine and revise entire pipelines conversationally with artifact support, moving beyond transformation-by-transformation edits to full pipeline-level iteration

Thanks for being part of the Nexadata journey!

— The Nexadata Team