AI in Project Management & ERP Implementation | Why Process Mapping Matters in Retail

AI can write your implementation plan. It can’t run your project (yet).

AI-generated project artefacts create the appearance of shared understanding.

The gap between appearance and reality doesn't surface until go-live pressure does, particularly across ERP implementation and retail project management environments.

It started with an "improved experience" email.

You know the ones - bright tone, cheerful copy and a promise that life on your platform is about to get easier.

The "improvement" was an AI bot on the front of the help site. And for simple questions, it was fine. I typed, it pointed, I moved on. Low stakes. No emotion. No need for pleasantries.

Then I hit a real problem. Complicated. Across configuration and data (and possibly the user having made some mistakes). Not a "click here, then there" solution.

I typed my question(s) and the bot gave me recommendations that were not quite related to my problem. I rephrased. Same result, different (and same) articles.

Five minutes in, I was not troubleshooting. I was trapped in a loop with a very confident, very useless gatekeeper sitting between me and someone who actually knew how to solve the problem - a pattern increasingly visible in AI-driven retail support environments, including ai in retail and broader artificial intelligence in retail industry applications.

When I finally broke through, I did not reach a seasoned specialist. I reached another layer - junior human, following their own script, contradicting advice I had already been given.

An AI gate. A junior layer behind it. The actual expert somewhere behind that.

Each "improvement" adding another buffer between the person with the problem and the person who could solve it.

I eventually found a workaround. Persist long enough and you could book directly with an expert. When I reached her, everything calmed down. The noise stopped. The problem got solved.

But it should not require a secret map and sheer bloody-mindedness to get to the person who knows what they are doing.

This Pattern Is Showing Up in Retail Project Management and ERP Implementation Too

What my CRM provider built is not unusual. It is, in fact, the logical endpoint of a very common approach to automation.

Layer tools over complexity rather than resolving it and call the result an improvement - something we continue to see across artificial intelligence in retail examples and emerging ai in retail examples.

In retail systems implementations, I see the same tendencies emerging across ERP implementation and retail system implementation programs. Project plans that look comprehensive because AI in project management tools and the PM drafted them in a fraction of the time.

Risk registers that cover every standard category but none of the specific ones that will land for this business. Status reports that are technically accurate and operationally meaningless.

Artefacts have never looked so polished. The thinking behind them? That’s what we want to preserve.

Brett Harned made this point directly in his analysis of project management in the age of AI: "If AI generates the plan before those conversations take place, the output might look impressive, but the project manager hasn't actually done the job."

For retail leaders running systems implementations, it is a more pointed warning than it might first appear.

Because in a project, the equivalent of my CRM's AI gate is a plan that looks finished.

The equivalent of the junior layer is a team that has been briefed but not aligned. And the equivalent of the buried expert is the actual decision-maker; the person who understands the business well enough to call it when the system is wrong - a capability grounded in strategic problem solving and supported by AI in decision making.

What Most Retail Leaders Get Wrong About ERP Implementation and Organisational Change Management

The assumption is that a better plan produces a better project.

So, leaders invest in the plan; the documentation, the process maps, the training decks, the communications schedule.

With AI generating competent first drafts of all of it, the investment feels even smarter, particularly when exploring how to use ai in project management and practical ai use cases in project management.

Here is what does not come with the polish: shared understanding of why the system works the way it does and who has authority when it does not - a core gap in organisational change management.

Prosci's research on ERP and systems transformation consistently identifies people and process failures (not technology failures) as the primary cause of implementation problems, including erp system implementation failures.

Specifically, the absence of aligned understanding across teams about decision ownership when things do not go to plan.

What makes this worse right now is the cognitive load paradox. When AI handles the administrative weight of a project, it frees up capacity.

But capacity freed from administration does not automatically become capacity for better thinking. Often it becomes capacity for fewer conversations.

"The systems are working as designed whilst the humans quietly check out."

Download the Thoughtful Automations DP

Fifty-five per cent of companies that made AI-driven workforce and operational decisions now regret them: not because the technology failed, but because they discovered too late what the humans had actually been contributing.

The institutional knowledge. The edge-case judgement. The relationships that held the system together when circumstances changed - often only recognised when ai implementation risk is underestimated.

Automating your project planning and execution does not eliminate these requirements. It hides them until go-live stress-tests them.

"AI can generate your implementation plan in minutes. It cannot generate the shared understanding that makes the plan executable."

How the Strongest Projects Approach Process Mapping and Business Process Mapping Differently

The retail implementations that actually stick; where teams hit go-live and keep building on the investment rather than quietly reverting to workarounds, share one defining characteristic. The project leader treats every artefact as a by product of conversation, not a substitute for it.

Process mapping is the clearest example. Most organisations want to skip it or compress it. It is unglamorous, it surfaces uncomfortable truths about how work actually flows versus how leadership believes it flows, and in a world where AI can generate business process mapping outputs quickly, it can feel redundant.

It is not redundant. Industry practitioners are direct about the prerequisite: "The process needs to be boring before you can automate it." The same logic applies before you implement a system to run it.

You need to understand the actual process (including the workarounds, the exceptions, the informal handoffs) before you ask a new system to execute it on a scale.

Research on automation maturity confirms that attempting to implement against processes that are not yet stable, standardised, or properly understood consistently results in higher failure rates, extended timelines, and inflated costs. The process mapping is not the deliverable. The conversations it forces are.

What the strongest project leaders do differently:

• They map current state processes with the people who actually do the work, not the people who designed them five years ago

• They mark decision points explicitly - who makes which call, under what conditions, and what happens when the system suggests something that does not look right

• They treat every working session as an opportunity to build shared understanding, not just to tick off a milestone

• They preserve human override so that teams experience the system as something supporting them rather than something happening to them

PMI's research on human-in-the-loop design confirms what experienced project leaders have known intuitively: the more complex the task, the greater the need for human intervention. A retail systems implementation is almost always complex.

That complexity does not reduce because the documentation is thorough.

"The real risk is not that your new system fails. It is that it succeeds whilst the people running it check out."

Your 3 Step Sanity Check for Project Planning and Execution

Before your next steering committee or project phase review, ask these three questions:

1. Do we have genuine agreement, or just silence?

Silence in a meeting is not alignment. The most expensive words in a project are "I thought we all understood that." If AI-generated summaries are substituting for discussion rather than capturing it, you are building a fragile plan (one that looks sound until the first real world complication tests it).

2. Who owns the decision when the system is wrong?

Every implementation will surface moments where the system does something unexpected - forecasts that miss a promotional period, rosters that do not account for leave, purchase orders triggered by logic that made sense at configuration and not six months later. If that question is not answered before go-live, you will answer it under pressure after it.

3. Are your team's workarounds mapped or hidden?

Harvard Business Review's 2026 analysis found that organisations are making systems bets on AI's potential, not its current performance. The same risk applies to process assumptions. If your team has undocumented workarounds in the current system, your new system will either automate them invisibly or break them loudly. Neither is a good outcome.

TL;DR

✓ The Hidden Mechanism: AI-generated project artefacts create the appearance of shared understanding without the conversations that produce the real thing and that gap does not surface until implementation pressure exposes it.

Diagnostic Questions:

• Can every team member explain why the system works the way it does, not just how to use it?

• Do you know which project decisions are documented versus genuinely agreed upon?

• If your project lead left tomorrow, would the team know who owns the hard calls?

Decision Framework: Before accepting any project artefact; plan, process map, risk register and ask: "Did this document come from a conversation, or did it replace one?" If you cannot answer that confidently, it is not ready to be signed off.

Ready to pressure-test your implementation approach before it costs you?

Book a 25-minute session with the 6R team.

Leonie McCarthy is the founder of 6R Retail, a systems implementation and change consultancy working shoulder to shoulder with Australian retail, wholesale, and manufacturing businesses to deliver projects that actually stick.

Retail improvement, made practical.
Leadership thinking that drives change.

Sign up to receive new articles and strategic guidance.