It’s great to know what you’re looking at and what’s happening, but it’s always better to get those views and context faster.

For instance, International Paper in Memphis, Tenn., recently started exploring how its 15 U.S. containerboard mills could migrate from traditional manual reporting and data movement to delivering context-driven and scalable insights. 

It began by conducting a trial of Seeq Intelligence software, including some of its most recent, artificial intelligence (AI)-based innovations, such as Agent Q, Agent Builder and Insight Boards to gain insights from its data more quickly, make them more easily available to its users, and scale them out. The trial consisted of an asset health monitoring project that included production areas, sub-areas, instrumentation and other components, and it used principal component analysis (PCA) inputs for anomaly detection.

“We didn’t have a problem. We had an opportunity because we’ve already been using Seeq Workbench software to create analyses, formulas and modeling, as well as reporting that requires data to be moved to Power BI, Organizer and Datalab software to carry out other reporting and presentation functions,” says Cameron Cox, data scientist on International Paper’s cost innovation and remote analytics team, which supports the company’s more than 400 manufacturing facilities and approximately 37,000 employees worldwide. Cox presented “Two paths to automated reporting: Insight Boards and Agent Builder reports” at Seeq’s Conneqt 2026 event earlier this year. “The opportunity was to keep further insights within Seeq, retain its trusted analyses, and perform fewer data movements, but still be able to provide different insights for different audiences without duplicating efforts, whether it’s looking at one mill or 15 plants.”

Insights from Agent Q

In the first part of its trial, Agent Q let International Paper leverage its process data with AI. It sits with other software agents as part of Seeq Intelligence, and lets users view additional insights within their data. More specifically, it helps with:

• Faster insight generation by using natural language prompts that enable rapid creation of signals and visuals.
• Understanding what matters, which helps identify useful KPI‘s, behaviors and conditions worth monitoring.
• Shaping reporting concepts that serve as a discovery tool for understanding what questions reporting should answer, in this case, about asset health monitoring.

Agent Q’s quickly generated, colorful bar charts show obvious performance concerns, but they also indicate the agent’s reasons for how it displays its graphics, which helps users with validation and understanding how its charts are created. As with most agents, Cox reports that Agent Q required specific details and thoughtful prompts—and practice developing them—to generate useful results.

“Agent Q is great for exploration and quick insights, and is flexible by letting users ask questions involving Seeq Workbench and other software tools,” adds Cox. “One of our process engineers told us, ‘By using Agent Q, I was able to find the root of less-controlled signals by letting it correlate the responses, and consider cascade delays. Think of it as fast-paced, root-cause analysis (RCA)-type feedback.’”

In essence, Agent Q lets users ask what’s happening in their process, or what caused productivity to decrease? However, while Agent Q’s explorations yielded valuable analyses and results, Cox adds that its responses were manual and inconsistent on different days, which showed that varied audiences needed suitable reporting formats.

To gather Agent Q’s outputs into consistent, repeatable, everything-in-one-place reports, Cox’s team implemented the Seeq Journal text documentation feature in Workbench, which can produce reports for today, tomorrow, 30 days or other periods. They also collected metadata from individual prompts in Journal, which enabled Agent Q to compile data, produce tailored reports, and relay them to predetermined Organizer software or other destinations.

Agent Builder reports in

In the trial’s second phase, Journal also allowed International Paper’s users to learn about and use Agent Builder, which allows reports to be generated according to certain times, events or other, more sophisticated triggers. Agent Builder and its nodes let Cox’s team automate the functions they developed in Journal, turn them into an agent, and get another step closer to fully automated reporting. Its primary functions include:

• Automating the agent’s reasoning process by creating multi-step logic to run analyses and assemble outputs automatically.
• Shifting from analysis to output with agents that can do the work and generate reporting artifacts.
• Learning that reliable reporting depends on repeatable execution paths and clear intents.

“Built from Workbench or Seeq Vantage analyses, Agent Builder uses plant-floor input to produce mill-level reports answering questions at that level, automates its multi-step execution procedure, and delivers outputs supporting prioritization to organizers, leaders and operators,” explains Cox. “These reports include an executive summary of alarms and other events over 30 days or other preset periods, as well as other insights and graphics, or AI-based recommendations about where to focus time and effort.”

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

Insight Boards show and tell

In the trial’s third phase, Cox’s team explored Insight Boards, which likewise enable operational visibility in three primarily graphical ways, including:

• Answering daily questions, such as how am I performing? What’s happening right now? And where should I look?
• Maintaining a continuous awareness design that’s built to detect outliers and anomalies by monitoring KPIs drifting over time, and identify emerging and urgent issues.
• Providing stable, trusted views, such as fixed layouts and saved visuals, which provide consistent, day-to-day interpretations, as well as the ability to generate quick graphics for one-off meetings, which can also be shared among users.