Preparing Your Ecommerce Team for Agentic Operations: A Change Management Guide

Agentic operations are workflows where one or more AI agents take on meaningful parts of ecommerce work, such as merchandising, customer service, email campaigns, or operational tasks, according to defined rules, approval points, and levels of autonomy. Instead of using AI only as an assistant for isolated tasks, agentic systems can move through several steps of a workflow: analyzing information, recommending actions, executing approved steps, and stopping when human review is needed.

For ecommerce brands, the benefits are tangible: faster execution, lower operational costs, and the ability to scale without scaling headcount at the same rate.

This is no longer theoretical. Shopify CEO Tobi Lütke made this expectation explicit in an April 2025 memo to his own team: before requesting new hires, teams must first demonstrate why AI can’t do the job. The question he asked managers to consider, "What would this area look like if autonomous AI agents were already part of the team?", is becoming relevant for every ecommerce brand.

Why Agentic Operations Requires a Different Kind of Change Management

Until now, change management in ecommerce meant helping teams adapt to new tools: migrating to Shopify, rolling out Klaviyo, or learning a new POS. The job stayed the same; only the instruments changed.

Agentic operations breaks that pattern. When AI agents start taking on meaningful parts of merchandising, customer service, or email campaigns, it’s not the tools that change, it’s the nature of the work itself. Teams go from doing the work to supervising the system that does it.

That demands a new kind of skill: supervising AI systems that can now perform parts of the work, while adjusting to a role that is changing in real time.

So what is a small ecommerce business to do?

Most guidance out there is written by enterprise consultancies for companies with 500 people and a Chief AI Officer. A 15-person Shopify team doesn’t have that, but the changes are hitting them just as hard. Nobody has the full playbook yet. Every organization is figuring this out in real time. But some things are already clear, starting with which roles are changing the most.

What Are the 4 Team Roles Most Affected by Agentic Operations?

Merchandisers

Merchandisers shift from manually curating collections and writing product descriptions to supervising AI agents that support or automate parts of that work. The core skill becomes quality assurance: reviewing outputs, catching errors, and refining the rules agents follow.

A merchandiser who once spent days building a seasonal collection may now spend that time reviewing, correcting, and refining what an agent prepared.

According to Shopify’s Q4 2025 earnings call, orders coming to Shopify stores from AI search increased 15x since January 2025, although Shopify noted this growth came from a small base. As AI-driven shopping scales, merchandisers need to ensure the product data, descriptions, and collections those systems pull from are accurate and well-maintained, which means more oversight is needed.

Customer Service Roles

When AI handles order tracking, returns, and basic product questions, the customer experience role moves up the complexity ladder. Agents become escalation specialists and AI quality reviewers, monitoring chatbot conversations and handling edge cases that require human judgment.

Email and Retention Marketing Roles

AI agents can now segment audiences, draft campaigns, and optimize send times without human input at each step. The email marketer’s role shifts from producing campaigns to designing the strategy and constraints the agent operates within, such as defining who gets what message and reviewing whether the output matches the brand.

The pattern in change management is consistent: when execution gets automated, the human role moves from player to coach.

Founders and Ecommerce Directors

For a 5-to-10-person Shopify brand, the founder or ecommerce director becomes the person making AI governance decisions. Their job now includes defining what the agents can and can’t do, who reviews outputs, and who owns the system when something goes wrong.

It also means deciding which workflows get automated, setting the constraints agents operate within, and maintaining quality standards across what they produce.

What Are the Most Common Resistance Patterns in Ecommerce AI Adoption?

Fear of becoming obsolete, lack of time to learn, and lack of trust in the technology: these are the resistance patterns that stall AI adoption on ecommerce teams.

Pattern 1: Fear of Becoming Obsolete (FOBO)

This is the deepest resistance and the most difficult one to address. Team members watch tasks they spent years mastering get done by a machine in seconds. The expertise that made them valuable suddenly feels irrelevant. Telling them “AI won’t replace you” doesn’t help because it doesn’t match what they’re experiencing. The fear of losing something always feels worse than the excitement of gaining something new.

To address this, leaders need to have the hard conversation honestly, then follow it by providing a concrete plan. It doesn’t need to be a perfect plan, but it needs to help the person gain a clear picture of what their role becomes.

Fear lives in ambiguity. If a CX agent knows their job is changing but doesn’t know what it’s changing into, they fill that gap with worst-case scenarios. Showing them specifically what their Tuesday morning looks like in the new model, such as reviewing AI conversations or handling escalations, replaces the fear with something tangible.

Pattern 2: No Time to Learn

Teams are already focused on their jobs. Asking them to learn new tools on top of their existing workload feels like one more thing on the to-do list. The resistance isn’t always about motivation; it’s also about capacity.

The solution comes from leadership: carving out protected time for learning, not expecting people to figure it out between tasks. A merchandiser isn’t going to explore AI-assisted collection management on a Friday afternoon after a full week of manual work. When leadership protects time for learning, it sends a clear message: this isn’t optional, and you’re not expected to figure it out alone.

Pattern 3: Lack of Trust in the Technology

Even when teams are willing and have the time, a third barrier often stops adoption: lack of trust in the technology to behave predictably. People worry that agents will take actions without approval, produce outputs that damage the brand, or make decisions no one asked them to make. The fear isn’t irrational. It’s a reasonable response to handing control to a system you don’t fully understand yet.

The answer is human-in-the-loop design. Every agentic workflow should have clear checkpoints where a human reviews, approves, or overrides before anything goes live. When teams can see exactly where they maintain control, and where the agent can and can’t act on its own, the technology stops feeling unpredictable and starts feeling manageable.

Trust doesn’t come from reassurance. It comes from architecture.

How Do You Actually Prepare a Team for Agentic Operations?

These barriers shape how any adoption effort needs to be designed. Here’s how we approach it at Lake House Group.

Step 1: Diagnose People’s AI Readiness

Not everyone on a team uses AI the same way or with the same level of proficiency. On a 15-person ecommerce team, it is important to assess where each person stands in terms of AI literacy, mindset, confidence, and readiness to work with agentic systems.

Without it, there’s no way to identify which workflows have the most potential or who is ready to work with them.

Step 2: Enable People to Understand the Tools

Once the tools are built, the team needs to understand them, not just how they work, but why they were designed that way. We don’t go into the technical details, but we do give an overview of the reasoning behind the tool’s logic: why the workflow runs in this order, why certain rules were set, and what trade-offs shaped those decisions.

When people understand the reasoning behind the tools, they stop feeling like the tools are being imposed on them and start feeling like they had a hand in shaping them.

We deliver this through collaborative workshops or just-in-time content that tackles the “we don’t have time to learn” pattern while creating a safe space for experimentation.

Step 3: Provide Ongoing Support

Agentic operations aren’t a one-time project. They evolve with technology, needs, and team skills. Ongoing support means continuous iteration on workflows alongside adaptive training that reinforces confidence over time.

Agentic operations are still new for most e-commerce teams. For smaller teams especially, the challenge is not to automate as much as possible, but to build systems that humans understand, supervise, and continuously improve over time.

FAQ

What is agentic operations?

Agentic operations are ecommerce workflows where AI agents take on meaningful parts of the work, such as merchandising, customer service, email campaigns, or operational tasks.

The shift is from using AI only as a task assistant to designing workflows where AI can analyze, recommend, execute approved steps, and pause for human review when needed.

Why does agentic operations require change management?

Because it changes jobs, not just tools. Previous technology shifts asked teams to learn a new interface. Agentic operations asks them to do a fundamentally different job, adapting mindsets, workflows, and skills.

Do employees need to be technical to work with AI agents?

No. Working with AI agents is less about technical skill and more about learning to supervise, review, and refine outputs. The shift is comparable to moving from individual contributor to manager: different skills, but not necessarily more complex or technical ones. What matters most is understanding what the agent is doing and knowing when to intervene.