Version 4.2: Scaling Industrial AI with DataOps + DevOps

The latest release of HighByte Intelligence Hub delivers critical capabilities for enterprise-grade Industrial DataOps. Version 4.2 represents another important milestone in our mission to continuously improve industrial data integration, contextualization, standardization, and scalability. 

In this release, you’ll find solutions for Artificial Intelligence (AI) on the factory floor, where DataOps supports AI and AI supports DataOps. This version also leverages DevOps tooling to manage Intelligence Hub deployments with the addition of Git and OpenTelemetry support. There are also new native connections with Databricks and TimescaleDB, as well as enhancements to many existing connections and functionalities. In this post, I’ll break down these key features and explain why they are so important for industrial data management. 

AI Meets Industrial DataOps

AI has massive potential for industrial companies when applied to their factory floor data due to its ability to process massive quantities of data, identify correlations and patterns, predict future outcomes, and provide specific data immediately when needed and requested.

However, there are challenges in taking industrial AI from theory to practical implementations. Organizations already struggle to access usable data at scale from the factory to feed consuming systems. AI exacerbates this challenge. AI has an insatiable appetite for high-resolution and high-quality data, and AI applications are highly focused and specialized, which results in many more AI agents and applications to be managed by OT and IT teams that are already resource-constrained.

The solution to these challenges is an industrial data strategy that uses an Industrial DataOps abstraction layer—like HighByte Intelligence Hub—for data orchestration, observability, governance, and quality. HighByte Intelligence Hub version 4.2 introduces new functionality that supports the relationship between Industrial DataOps and AI.

• DataOps for AI provides the underlying framework that connects, organizes, and prepares data, providing a reliable data foundation for successful AI deployment.
• AI for DataOps improves the efficiency and scalability of DataOps practices—automating the process of gathering, contextualizing, and preparing data from various sources for analysis.

Industrial DataOps for AI

Agentic AI is the next evolution of AI, enabling the deployment of autonomous “agents” that can perform advanced tasks. AI agents are standardizing on Model Context Protocol (MCP) to communicate with systems, understand what data is available, and access this data. 

In version 4.2, the Intelligence Hub includes the first Industrial MCP Server that allows you to build custom MCP tools for agents to interact with industrial systems. Build a tool that allows AI agents to interact with the MES system, CMMS, or operational data in files. You have complete control over the scope of what the AI agents can see and do, and you can rapidly iterate over tooling to improve their efficiency and outcomes. The tools are not static, and they are completely in your control. In a world where AI agents will have specific tasks and must be permissioned for specific tasks, this kind of control at the data layer is critical.

For example, you can build a tool called “CreateMaintenanceWorkOrder” that an AI agent can call when it decides a machine needs maintenance. Using a pipeline in the Intelligence Hub, this tool can validate the AI agent’s input, create a work order in Maximo or MaintainX, and then respond to the agent with the work order number.

There are numerous opportunities for AI agents. Use the Industrial MCP Server for HighByte Intelligence Hub to source data in real time from multiple systems, make real-time or historical data requests on these systems, consolidate, standardize and contextualize data through models, and execute multi-step processes through pipelines. See the Industrial MCP Server in action.

AI for Industrial DataOps

AI can also be used in service of Industrial DataOps. New in version 4.2, use AI to scale out software configuration in the Intelligence Hub. 

When curating data from large namespaces like a factory-wide OPC server, there may be millions of tags and thousands of individual assets that need to be identified and contextualized. In version 4.2, the Intelligence Hub has AI-assisted instance contextualization utilizing new connections to Amazon Bedrock, Azure OpenAI, Google Gemini, and OpenAI, and the ability to connect to local LLMs. 

In this new contextualization process, you can define an example mapping of an instance for a machine from an OPC UA address space. The LLM is passed this instance definition and the OPC UA address space, and then the LLM scans the address space to identify and propose similar instances. You have the option to create or reject the AI-identified instances, bypassing the time-consuming task of manually identifying machines and mapping instances. The result is a significant reduction in the overall work required to leverage telemetry data for digital projects, particularly in large brownfield facilities. 

To learn more about accelerated tag mapping in the Intelligence Hub using AI, watch this video by HighByte CTO Aron Semle.

DevOps for DataOps

Deploying secure and scalable software integrations across an enterprise requires deliberate governance and management of change. At the same time, large deployments require real-time telemetry from software applications to monitor each running software deployment. 

To meet the governance and observability demands of the industrial enterprise ecosystem, HighByte Intelligence Hub now supports Git for governance and version management, as well as OpenTelemetry for software monitoring at scale. 

Advanced version management for data models and pipelines

Scaling software deployments across a single site is challenging and becomes even more difficult when done across multiple sites. Projects are implemented, tested, and deployed and then more use cases are added. As that happens, production equipment is changed, and data requirements evolve. This requires version management of the configuration and approval workflows to ensure that production system data collection is reliable, consistent, and robust. 

The Intelligence Hub now integrates with Git, allowing administrators to backup configuration to Git and to pull the latest Git-approved versions from their repos when deploying HighByte Intelligence Hub containers.

DevOps teams commonly use Git to manage change processes, approved versions, and deployment of configuration files and scripts. HighByte Intelligence Hub version 4.2 adds support for automating the backup and deployment of an Intelligence Hub instance from one or more Git repositories, streamlining DevOps workflows. Upon deployment, the hub supports pulling down one or more repositories and merging and launching the configuration. This allows for deployments that have both common enterprise configuration (i.e., models, pipelines) and custom site configuration (i.e., instances, plant connections).

Enterprise-grade observability with OpenTelemetry

In addition to having the right version of the software and configuration files, large deployments also need real-time telemetry from software applications to monitor and alarm on each running deployment. Most IT departments monitor the many applications they support with a single observability solution. 

OpenTelemetry is an open-source observability framework that standardizes the collection of metrics, logs, and traces from applications. In version 4.2, the Intelligence Hub now supports publishing metrics and logs over OpenTelemetry to observability applications.   

OpenTelemetry is supported by many IT-based monitoring tools, including Dynatrace, Datadog, Honeycomb, Splunk, AWS CloudWatch, Azure Monitor, Prometheus, Grafana, and more. By adding OpenTelemetry support, the Intelligence Hub now makes system metrics like CPU and memory—as well as application metrics like pipeline and connection performance—easily available to any of these tools.

Enhanced Connections

Databricks

Databricks is a widely adopted enterprise data platform for data science, analytics, AI/ML, and lakehouse solutions. While HighByte Intelligence Hub users have historically integrated with Databricks through cloud object storage, the limitations became clear, and a native connection became an imperative to improve the experience between the two platforms. 

Version 4.2 of the Intelligence Hub introduces a new native connection to Databricks. The Databricks Client connector publishes optimal, ready-to-use data in parquet format to the underlying storage service and registers its contents with the Databricks Unity Catalog. This allows you to dynamically drive your lakehouse schema from operations and instantly discover and use your industrial data.

TimescaleDB

HighByte Intelligence Hub version 4.2 introduces a new connection for TimescaleDB. Built on top of PostgreSQL, TimescaleDB is familiar for relational data and intuitive to learn for time series data. This new connection’s outputs provide the option to create hypertables for time series datasets such as asset telemetry and sensor data, as well as standard tables for relational datasets such as location identifiers, asset metadata, MES transactions, and QMS results.

Also new in connections

HighByte Intelligence Hub version 4.2 includes several enhancements to existing connections:

• The Oracle Database connection now supports Change Data Capture (CDC).
• The Snowflake SQL connection now supports outputs, including Insert, Upsert, and Update commands. This improvement makes it easier to perform SQL-like operations on data in Snowflake.
• The Amazon S3 and Apache Kafka connections now have input support to read data out of S3 and Kafka to make this data available at the edge or to other cloud services.

Wrap Up

In the pursuit of industrial AI-readiness, DevOps application management, and deeper industrial connectivity, version 4.2 of the Intelligence Hub takes a big step forward. Those looking to accomplish Industry 4.0 use cases and enterprise-level data integration can rely on HighByte Intelligence Hub as best-of-breed technology for the transformation and use of industrial data in modern applications.

To learn more, please check out these additional resources:

• Read the product announcement
• Explore the release notes for details on all new features and fixes
• Request a free trial or log in to your existing account to test and deploy the software in your unique environment