< Back to Blogs
Share:

Hire, Onboard, Manage: Treating AI Agents Like New Team Members

Special Guest Blog Post written by Deborah Dull 

 

Imagine it’s launch season in your supply chain, and a team of AI agents is hard at work: coordinating shipments with suppliers, allocating inventory, and handling a missed delivery.

It’s hard to go a day without hearing about the latest AI headlines: new LLMs, new applications, new use cases. These discussions are part of my every day. Business leaders and IT teams are eager to explore – and they have a lot of questions… and apprehension.

In these discussions, we have found one framework that lands consistently: consider AI agents like new hires, not new software.

 

Step 1: Hire Like You Mean It

Just like hiring humans, this step is about defining what you need. The first decision is the purpose of the role: what problems are you trying to solve? Where are your people overwhelmed? Where is your business growing in coming months?

From here, the next decision is to build the business case just like you would for a new headcount. For example, consider the ROI of having a digital employee who can process supplier performance data 24/7.

Now, decide on the type of agent that will join you team. You might hire an embedded AI from your existing software providers such as ServiceNow or Kinaxis, specialized agents that perform a single task such as from Zip for procurement or Daybreak for forecasting, or full agentic workforces such as from Qurrent or Auxo.

The last decision is to “interview” your candidates and select the one that will be the best fit technically and culturally.

 

Step 2: Onboard for Success

Just like that nervous new hire walking through your office doors on their first Monday, your AI agent needs to go through an onboarding step. The difference between digital and human workers is that your digital employee learns faster, remembers everything, and never needs coffee breaks.

The first onboarding phase lasts about two weeks. Your AI agent needs system access and basic training. For example, an overview of the process to manage supplier performance. The result feels like an overeager intern, fresh to the workplace and unaware of the way work “really” gets done.

The magic really happens in the next phase. In this phase, your human workers give feedback to your digital workers, just like they would give feedback to a human. “This supplier has struggled with on time performance but has recently committed to improve. Be strict on follow-ups but friendly on tone” has immediate influence on how the agent interacts with that supplier. But unlike humans who might forget the details of a supplier issue from six months ago (or go on vacation!), AI agents remember everything. The overeager intern becomes an expert in just a few weeks.

 

Step 3: Manage Your Hybrid Workforce

This step is all about performance management activities. The first activity is setting expectations. For example, when a distribution company deployed an AI agent to handle their returns process, they established clear performance standards: process 95% of return requests within four hours, maintain customer satisfaction scores above 4.5, and escalate complex issues to human staff within defined parameters.

The next activity is supervision. Since AI agents work around the clock, their “first line supervisors” do also. This activity is about having agents watching agents, with clear instructions on the type of performance to manage. Humans provide final oversight and handle exceptions.

Another activity is managing human workers. Managing the performance of a hybrid team feels a little different for the human workers. As “run business” and on-the-fly issue resolution moves to AI agents, humans can focus their time on structured problem solving and relationship building. This shifts the performance management time horizon into longer term goals and results. This means a reduction in the daily fire-fighting dopamine hits, replaced by the delayed satisfaction of longer-term wins. For some, this is a welcome break. For others, they’ll need help through the transition.

Ready to Start Hiring?

The stepwise change of Agentic AI in the workplace is happening. Following these steps to lead your supply chain into the new age of hybrid teams.

Ready to dive deeper? “Self-Driving Supply Chain: 5 Steps to Building Your Agentic AI Workforce” is launching later this year and provides the framework for welcoming digital workers into your supply chain

 

Deborah Dull is Managing Partner of Trillium Digital Services leading Agentic Services. With experience at GE, Microsoft, and Zero100, she specializes in AI-driven supply chain innovation. Her upcoming book “Self-Driving Supply Chain” explores agentic AI workforce implementation in supply chain operations.

More Blogs

warranty management optimization
Blogs
May 5, 2026

Can the Warranty Claims Process be Improved?

written by Chris Cunnane with InterSystems   Automotive manufacturers have invested billions in digital transformation over the past two decades, achieving meaningful gains in production efficiency. Yet one area has stubbornly lagged behind: warranty cost performance. Despite advances aligned with Industry 4.0, warranty costs as a percentage of revenue have remained largely unchanged. This disconnect highlights a critical truth that while production has evolved, warranty operations have not kept pace. At the heart of the issue is a fundamental inefficiency: time. Specifically, the lag between when a warranty claim is submitted and when meaningful problem-solving begins. This delay not only drives up operational costs but also impacts customer satisfaction and brand perception.   The Hidden Complexity of Warranty Claims Warranty management is more than a back-office function. It is a complex, data-driven process involving multiple stakeholders, systems, and validation steps. Each claim requires verification of warranty coverage, assessment of the issue, and coordination across service providers, suppliers, and internal teams. Much of this process remains manual. Organizations often rely on fragmented systems, inconsistent data formats, and time-intensive validation steps such as reviewing receipts, maintenance records, and warranty agreements. These inefficiencies introduce delays that can stretch into weeks before root cause…
demand sensing
Blogs
March 3, 2026

Key Demand Sensing and Forecasting Use Cases Across Industries

Special Guest Blog Post written by Chris Cunnane with InterSystems   In a world defined by rapid market shifts, volatile supply chains, and unpredictable customer behavior, traditional forecasting methods often fall short. Relying primarily on historical data is no longer enough. To stay competitive, organizations are increasingly turning to demand sensing and forecasting, an approach that blends real-time data, advanced analytics, and AI to anticipate demand more accurately and respond faster to change. This shift is not limited to retail or manufacturing. Demand sensing is transforming how organizations across industries plan operations, allocate resources, and improve service levels. Below, we explore key industry use cases where demand sensing is delivering measurable value, and why businesses should care.   Why Demand Sensing Matters Demand sensing moves beyond static historical trends. It incorporates current, high-velocity data signals such as sales transactions, weather patterns, logistics feeds, economic indicators, and even social sentiment to generate short-term demand forecasts that reflect real-world conditions. The benefit is clear. Organizations gain better visibility and responsiveness across procurement, production, inventory, and distribution. Instead of reacting to outdated forecasts, they can make timely decisions that reduce costs, prevent stockouts, and improve customer satisfaction. FMCG, CPG, Retail, & E-Commerce Fast-moving…