Building an AI-Ready Data Foundation

Modernizing Industrial Data Architectures, GxP Compliance, and IT/OT Convergence in Life Sciences

Axendia participated in HighByte’s DataOps Day Boston 2026 at the Cambridge Innovation Center in Cambridge, MA. Building upon insights from our previous HighByte briefings, this event brought together life sciences manufacturers, automation specialists, and leading technology and integration partners, including Siemens, Snowflake, AWS, Tulip, and Skellig.

The overarching theme of the day was clear: as manufacturers transition from digital transformation pilots to enterprise-scale Industrial AI, success hinges on establishing a contextualized, governed, and validation-ready data foundation.

Key organizational imperatives voiced throughout the event included:

  • Early Cross-Functional Alignment: IT, OT, and Quality need to be in the same room when deploying digital systems that affect them all. The sooner the better.
  • Human-Centric Implementation: When in doubt, talk to your counterparts – the people that will be involved with the implementation.
  • Pragmatic Scaling: Don’t try to boil the ocean. Start small and stay steadfast. Get an executive champion. Measure the impact and use that data to justify the next program.

HighByte Intelligence Hub: From Unified Namespace (UNS) to the Data Control Plane

HighByte Chief Product Officer, John Harrington, and Chief Technology Officer, Aron Semle, highlighted how Industrial DataOps has evolved. While the Unified Namespace (UNS) established an architecture for standardizing and democratizing telemetry, manufacturers scaling across enterprise sites require a Data Control Plane to manage data volume, contextualize data products, enforce data contracts, and govern access for AI agents.

With over 120 global customers—including life sciences companies such as Regeneron, Bayer, Alcon, and Catalent—HighByte demonstrated how HighByte Intelligence Hub decouples hardware and software. According to Harrington, moving from point-to-point integrations to factory APIs and solution patterns accelerates data project deployment up to 10X while maintaining data integrity.

Source: HighByte

Reflecting on HighByte’s solutions, Semle noted, “The team’s expertise in Industrial DataOps has helped us stay focused. It’s easy to get pulled in different directions, especially with the obsession around AI right now. Interestingly, AI has reinforced the need for contextualization, so we find we don’t need to push that concept as much as we used to.” Semle continued, “AI is lowering the barrier between problems and solutions. Frontline workers, with no experience in coding, HMI, SCADA, etc. can now develop applications that solve their specific problems.”

Source: HighByte

HighByte’s Director of Product Management, Jeffrey Schroeder, introduced key capabilities in the Version 4.5 release, including agentic configuration, modeling agents, lookup maps for cleaning telemetry, and expanded data lineage features designed to support internal and external agent workflows safely.

Real-World Case Study: AI-Driven Alarm Management at Scale

Rory Sheehan, Global Head of Sales at Deep Digital Solutions Group, presented a case study on alarm rationalization at a pharmaceutical manufacturing company. Sheehan explained that plant operators faced almost 88 alarms per hour, leading to severe operator overload. By linking raw alarm history from SCADA and DCS systems with batch records, ISA-88/ISA-95 equipment hierarchies, process states, and shift patterns, the team identified a quiet, un-contextualized agitator seal low-pressure alarm. According to Sheehan, this single unflagged alarm had stalled 8 batches for a total of 31 days.

Sheehan explained that, by establishing contextualized data products, the plant expects to eliminate ~124 days per year of stalled production time in a single facility, while reducing alarm investigation and review time by over 40%. Sheehan highlighted that the system strictly separates deterministic queries from probabilistic AI outputs, ensuring human-in-the-loop oversight for GxP operations, and he emphasized that AI initiatives must be rooted in solid data infrastructure. He reinforced that this is a data project (not just an “AI project”), with AI only as good as the context underneath it.

Customer Spotlight: Regeneron’s Modern Industrial Data Architecture

Ronald Sherrod, Staff Automation Engineer at Regeneron, shared how his team built an event-driven, scalable data spine to replace legacy point-to-point connections.

Pointing out the flaw of legacy architectures, Sherrod noted that every new system adds yet another set of interfaces – each one designed, tested, validated, and maintained on its own – which is not scalable.

To overcome this, Regeneron deployed their “Plant Bus” hub-and-spoke reference architecture combining Ignition gateways, HiveMQ Enterprise MQTT brokers, and HighByte Intelligence Hub. This setup enables report-by-exception publishing, modeling context once at the edge, and reusing data products across analytics, MES, and enterprise platforms.

Regeneron’s architecture embeds ALCOA+ data integrity principles directly into the data layer, utilizing store-and-forward capabilities, central configuration, and audit trails. By applying a Computer Software Assurance (CSA) risk-based approach, Regeneron validates underlying data modeling patterns once. Subsequent site or asset integrations become governed configuration changes rather than full system re-validations.

Based on his team’s lessons learned, Sherrod offered six key recommendations for industry peers:

  • Start with the use case: Agree on success criteria before the first connection.
  • Bring Quality in early: Treat QA and CSV teams as design partners. A shared understanding of the intended use will save months.
  • Govern models like code: Each needs to be versioned, reviewed, and promoted through environments under change control.
  • Design for failure: Assume outages and duplicates. Asynchronous “store and forward” and handlers that safely handle repeated messages (idempotent) are not optional.
  • Stay standards-based: Open standards like MQTT, Sparkplug B, and OPC UA keep each layer replaceable.
  • Deliver value in slices: Prove one high-value end-to-end flow, then scale the pattern.
Source: HighByte

Panel & Ecosystem Highlights: Modernizing GxP Compliance

During the panel discussion, Lessons From the Field: Modernizing GxP Compliance, moderated by HighByte CCO, Torey Penrod-Cambra, panelists Ashtad Engineer (AWS), Amy Williams (Skellig), Dillon Forzese (Tulip Interfaces), and Ronald Sherrod (Regeneron) discussed practical strategies for digital transformation:

The Modersys team has captured a set of representative operational outcomes from its customer base: 

  • Speed of Deployment: Forzese shared how establishing standard data products enabled Tulip to deploy a solution at one site in two months, followed by rapid scaling across 17 additional sites in just two months.
  • Quality & Governance: Panelists agreed that involving Quality early transforms compliance from a roadblock into an enabler. As one panelist noted, “If you’re innovating correctly, your quality improves.”
  • Proving Business Value: Williams emphasized that securing executive buy-in requires focusing on clear operational pain points, while Engineer noted that eliminating manual transcription directly answers the ROI question.
Source: HighByte

Siemens Industrial Edge & Intelligence Center X

Following Siemens’ partnership announcement with HighByte in June 2026, demonstrations at the event showcased how Siemens Industrial Edge, HighByte, and Siemens Intelligence Center X operate in unison. HighByte serves as the critical contextualization layer bridging shop-floor OT (PLCs, SCADA) with cloud analytics platforms like Snowflake and AWS. Troy Steen, VP of Siemens Corporate Development, highlighted this value: “This enables Knowledge Graphs and AI Studio to deliver role-based, actionable insights to shop-floor personnel and executives. The HighByte partnership fills a gap, enabling the shop floor leader to act on alerts and surface that information to other teams more quickly.” Steen noted that other high-value use cases include predictive maintenance, managing manufacturing quality escapes, and mitigating supply chain disruptions driven by tariff and delivery uncertainties.

In Brief

As Life Sciences organizations accelerate the adoption of Industrial AI and agentic workflows, the distinction between raw data transport and contextualized data control becomes paramount.

A UNS or Data Control Plane is the AI-enabling infrastructure, not AI itself. It drastically reduces the complexity and cost of delivering structured, real-time data to models. However, maintaining GxP compliance, data integrity, and operational trust requires rigorous governance, clear systems of record, and an unwavering commitment to keeping humans in the loop for critical decision-making.

We will continue to provide updates on HighByte, Industrial DataOps, and Life Sciences digital manufacturing trends as they become available.

Ready to unlock new opportunities and drive success? Schedule an Analyst Inquiry 

Related Content

Stay Connected with Axendia!

Stay at the forefront of innovation and technology in life sciences, healthcare, and space-enabled research by connecting with Axendia.

  • Follow us on LinkedIn: Join our professional network and stay updated on the latest insights, trends, and case studies. Connect with Axendia on LinkedIn
  • Subscribe to Our Updates: Get exclusive insights and thought leadership articles delivered straight to your inbox. Sign up for our newsletter
  • Share Your Thoughts: We’d love to hear from you! Contact us to discuss your ideas, challenges, or opportunities. Email Axendia

Let’s drive innovation together!


The opinions and analysis expressed in this post reflect the judgment of Axendia at the time of publication and are subject to change without notice. Information contained in this post is current as of publication date. Information cited is not warranted by Axendia but has been obtained through a valid research methodology. This post is not intended to endorse any company or product and should not be attributed as such.

Scroll to Top
Share via
Copy link