2026 Forum Reflections

How Avantor and Aera Technology Are Operationalizing Decision Intelligence: Insights from ARC Advisory Group’s 30th Leadership Forum

During the 30th Annual ARC Advisory Forum on February 10, my session, “The New Frontier of Operations and Supply Chain,” was a 1.5-hour discussion filled with rich learning opportunities. The session served as a space for professionals to share their experiences and insights into the future of supply chains. Key highlights included real-world end user case stories, high-level discussions on implementing novel solutions, and integrating AI into operational processes.

Jared Guckenberger was the end user presenter during the session, showcasing the results of implementing Aera Technology’s Decision Intelligence solution. Jared is the VP of Global Supply Chain at Avantor. Avantor provides mission-critical materials and tools to life science companies, biopharmaceutical producers, and medical R&D organizations. Avantor has a global reach across 175 countries, 40 distribution centers, and various closer storage sites.

Scale:

• 10,000 supplier/source combinations.

• 250,000 SKUs sold per year.

• 1.5 million SKU-location combinations.

• 10M+ purchase and customer orders per year.

Jared shared Avantor’s supply chain challenges, including extremely high transaction and data volumes. Inventory challenges included having too much, too old, and too little stock at the same time (excess, write-offs, and stockouts). He emphasized the need to sense, decide, and act faster, as well as to integrate better with suppliers, many of whom are low-tech and non-EDI. Many solutions also need to be change-ready, scalable, and usable by multiple roles, rather than relying on fully autonomous AI.

Jared played a pivotal role in establishing a working relationship with Aera Technology to address the company’s supply chain challenges using Aera’s Decision Intelligence solution. This approach is not only about agentic AI but also includes classic machine learning and decision logic, all orchestrated into repeatable decision processes. The core idea of Decision Intelligence is to enable the system to make thousands of routine decisions that humans do not have time for, rather than attempting to make decisions that are “smarter than humans.”

At the start, Avantor focused on three targeted skill areas, concentrating on decisions and processes that were traditionally inefficient. The company prioritized stock rebalancing, purchase order cancellation, and purchase order prioritization to address inventory issues and improve customer service.

Stock Rebalancing: In the past, this was largely a manual, monthly exercise with significant operational churn, and distribution centers were reluctant to spend days loading trucks, so only top items were prioritized. Now, the system scans twice a week to identify dead or slow-moving stock and target locations with demand. Decision Intelligence generates move recommendations, which planners can approve or reject. Once approved, stock transport orders are created automatically in SAP. The overall impact includes a shift from infrequent batch rebalancing to continuous, every-other-day rebalancing. This approach captures many small opportunities that humans previously ignored and helps reduce write-offs and dead stock.

Purchase Order Cancellation: Traditionally, dynamic demand changes (orders canceled or modified) resulted in slow response times of two to three weeks. Now, the system scans weekly and proposes PO cancellations. Recommendations are sent to suppliers via email, without requiring EDI. Suppliers reply by email, and the system parses and summarizes the responses. The buyer then decides whether to accept or deny the cancellation. This process has reduced cycle time from weeks to about a week or less. In the early phase, this approach has already saved approximately $300K in inbound POs within one to two weeks using a small group.

Purchase Order Prioritization: Avantor has transitioned from a reactive to a proactive approach to managing stockouts. The system now predicts potential stock shortages based on current demand and supply data. It automatically emails vendors with requests such as moving deliveries forward by a specified number of days. Vendor responses (yes, no, or partial) are processed by the system and presented to the buyer, who confirms the changes after considering associated costs. This proactive service enhancement improves the customer experience without relying on Advanced Planning and Scheduling tools such as SAP APS.

Key Lessons

Don’t wait for perfection; go live and then iterate. Avantor currently has 62 enhancements still in the backlog. Organizations should select use cases with clear, immediate business impact and strong business sponsorship. It is also important to simplify processes where possible before or alongside automation. Finally, having a roadmap is essential, as successful pilots quickly create demand for additional AI and decision intelligence use cases that must be prioritized and funded.

Executive Leadership Q&A Panel

After the presentation portion of the session, the discussion moved into a Q&A format with various industry professionals, including:

• Peter Quimby of Avantor.

• Jeremy Hudson of Open Sky Group.

• Bryan Batchelder of Datex.

• Jared Guckenberger of Avantor.

With over an hour of discussion, the following were some of the key questions and responses from the panelists.

Question 1: As a system integrator, what best practices should customers follow when integrating new solutions, especially around data and AI?

Jeremy Hudson noted that many prospects try to tackle the most complex use cases first or attempt to replicate high-profile keynote examples such as digital twins and robotics. He recommended starting with simpler, high-impact problems, especially where not all partners provide ASNs or EDI. The focus should be on supporting decision makers, accepting that data will not be perfect, and selecting tools that can work effectively with imperfect data.

Question 2: How do you handle a go-live when the system is not perfect yet? Did you run a parallel system, and how did you manage the risks?

Jared explained that there was no parallel legacy system because they effectively started from a baseline with limited existing systems, making a head-to-head comparison impractical. He also mentioned the creation of a steering team to communicate extensively with the distribution network, including explaining trade-offs such as spending $15 on expedited shipping to avoid losing $1,000 in inventory value.

Peter Quimby added that Decision Intelligence is fundamentally about explicitly designing, evaluating, and learning from decisions. Organizations must accept some early friction to begin capturing valuable learning signals.

Question 3: Bryan, your role at Datex combines back-end development and front end product management. What capabilities have you been working on recently, and how have customers responded?

Bryan described how the Datex platform is built on a low-code application framework that enables professional services teams to implement customizations more quickly and cost-effectively. He also highlighted ongoing work on embedding AI and agentic coding tools to help users define data sources, essentially structured queries over operational data, which can then be wrapped into reports. In addition, the company is working toward a multi-agent orchestration and execution environment, where customers could deploy agents loaded with contextual data ranging from sales prospecting to post-implementation insights.

Final Thoughts

The session highlighted the current thinking and actions of leading companies in the supply chain market. A paradoxical trend is emerging: disruptions are increasing while organizations are simultaneously achieving new levels of operational efficiency. Both end users and solution providers are navigating distinct challenges, yet they share common goals of improving resilience, enhancing efficiency, and maintaining a competitive edge through digital transformation.