Data-source links let LLMs understand and deliver

To illustrate virtualization and AI’s potential flexibility, Kudzai Manditereza, senior industry solution advocate at Hive MQ, reports it can be helpful to think about reconfiguring a filling or packaging machine from processing 1 liter bottles to running 1.5 liter bottles.

“Shifts like this typically require many changes in materials and settings. Previously, users had to accept living with hardcoded devices that didn’t allow more than minimal changes,” says Manditereza. “Now, system virtualization and AI allow much greater flexibility, and easier and quicker reconfigurations. They also let users adopt tiered approaches with containerized software at the edge-computing layer.”

To connect legacy PLCs and interfaces, pull data, and integrate with higher-level systems, HiveMQ Edge software can run virtually on Level 2 edge servers. It can also convert production data to new protocols, and relay it from a plant’s IT layer via MQTT protocol to HiveMQ Enterprise broker and HiveMQ Pulse software, which coordinate to take typically raw data tags, add business context, display results, and close the loop by sending subsequent instructions back to the plant and its operators.

While it doesn’t run its own internal AI software, HiveMQ connects data sources and processing platforms, so AI and other large-language models (LLM) can make sense of it, and deliver useful insights to users. “No one wants to let AI take action without oversight, and users also want to validate data first to make sure they can trust its lineage,” explains Manditereza. “So the question is, who to partner with for AI? Traditional AI, machine learning (ML) and predictive evaluations are available, but users also want to employ generative AI (gen AI) with their production systems and enterprise applications, so they have to go through a layer like HiveMQ. To provide a consistent model on top of data streaming, we also added a data backbone last year called HiveMQ Pulse for distributed data intelligence.”

Platform provides information across layers

For example, Eli Lilly & Co. identified a connectivity gap across its lab and manufacturing facilities in 2022, including many standalone benchtop instruments such as pH meters and balances. They weren’t integrated with any computer systems, which could lead to potential regulatory compliance challenges.

To close this gap, the company started a cross-functional effort by its IT, engineering and quality departments to achieve cloud-enabled equipment connectivity. Its goals were automating data collection, improving data integrity and advancing Lilly’s digital transformation. Consequently, the teams developed Lilly’s Equipment Connectivity Platform to provide a standardized, scalable interface layer for connecting laboratory and manufacturing equipment to centralized systems, such as manufacturing execution systems (MES) and laboratory execution systems (LES), while also transmitting data to the cloud. Built-in controls ensure data compliance, and comprehensive reporting capabilities support thorough data review and accountability.

Eli Lilly reports it takes advantage of HiveMQ’s standardized, scalable interface to MESs and LESs, which includes:

• Faster integration with MESs and LESs;
• Enhanced equipment interoperability across multiple systems
• Technical controls that prevent or detect unauthorized testing and unreported data;
• Automated content capture and flows, reducing manual data handling;
• Scalable, flexible cloud-based technology for seamless operations, and
• Standardization across multiple sites, reducing paper processes and further streamlining workflows.

Get your subscription to Control's tri-weekly newsletter.

“Eli Lilly previously used USB sticks to manually pull data from spreadsheets, send it to IT, and make it available to users. This was a long chain with many potential points for errors, and it was hard for the company to consolidate its information,” adds Manditereza. “To unify its data, and make it accessible across layers throughout its enterprise, the company needed to include all sources and types in a framework, which decouples IT and OT organizational structures, and makes information more accessible to enterprise-level software. HiveMQ Broker connected Eli Lilly’s edge-layer software in an open architecture, used MQTT to reach its PLCs and their data, incentive to HiveMQ Broker using one protocol, layer and language. HiveMQ Pulse is the tool for building unified namespaces (UNS).”

The pharmaceutical manufacturer’s updated connectivity and new capabilities rely on HiveMQ, which serves as the core MQTT hub for secure, real-time data exchange. Instruments in the labs and manufacturing areas publish data through MQTT topics using Sparkplug B standard protocol for Industrial Internet of Things (IIoT) applications. The company’s MESs and LESs subscribe to this data, which is seamlessly transmitted to a cloud-computing service using HiveMQ’s bridge configuration and Kafka extensions. Next, HiveMQ’s centralized message broker provides a consistent security model, ensuring safe and efficient data transfers between equipment and the cloud. This capability is critical for maintaining compliance, supporting real-time decision-making, and enabling operational efficiencies across Lilly’s plants. The Equipment Connectivity Platform also empowers staff to leverage a scalable, edge-compute solution, and transform operations by letting its processes run more productively.

In addition, Lilly operates multiple HiveMQ licenses for its enterprise-server and production sites. So far, its Equipment Connectivity Platform powered by HiveMQ has connected several hundred instruments across its laboratory and manufacturing sites. The company was expected to expand this platform to more sites during 2025. The platform presently incorporates HiveMQ’s Security and Bridge extensions, while the company plans to leverage HiveMQ’s Data Hub and Data Lake extensions as part of future Edge 2.0 UNS implementations. Given the platform’s success, Lilly also formalized this architecture into a global connectivity strategy that aligns with its digital plant vision for fully connected facilities.