ERP Consulting and Microsoft Dynamics 365 Business Central Implementation in the Age of AI Programming: Why Experienced Consultants Matter More Than Ever

By Rajib Lochan Huzuri

GitHub Copilot, Claude Code, Cursor, ChatGPT, Microsoft Copilot, over the past couple of years, these tools have changed the rhythm of software development more than anything I’ve seen since the shift to cloud-based ERP. Developers generate code snippets in seconds, draft documentation without complaint, and run through testing scenarios that used to eat entire afternoons. It’s genuinely impressive, and I use these tools myself, daily.

But a Microsoft Dynamics 365 Business Central implementation isn’t really a software development project. It’s a business transformation project that happens to involve software. Miss that distinction, and you’ll misjudge what AI can and can’t do for you.

ERP Is About Business Transformation, Not Software Installation

Technology is the enabler in an ERP project, not the point of it. The implementations I’d call genuinely successful are the ones where leadership understood, early on, that they were signing up for more than a system swap. They were redesigning how work actually gets done, cleaning up processes that had accumulated years of workarounds, tightening financial governance, getting finance and operations to agree on shared numbers instead of competing spreadsheets, and giving executives data they could actually trust when making decisions.

I think of a manufacturing client who came to us wanting to “automate inventory.” Fair enough, that’s what they thought they needed. But three weeks into discovery, the real project turned out to be procurement policy: how they selected suppliers, how they negotiated terms, how much working capital they were tying up in safety stock nobody had reviewed in years. Business Central became the vehicle for that redesign, not the reason for it. That’s transformation. It isn’t coding, and no AI assistant, however capable, is going to have that conversation with a CFO about why the company’s ordering habits don’t match its cash position.

Why Business Central Projects Still Need Experienced Consultants

AI can write AL code competently. It cannot run a discovery workshop and notice that the CFO’s answer to a question wasn’t really an answer, it was a hint about a problem nobody had named yet. That’s where experience earns its keep, across a set of activities that show up in nearly every implementation:

Requirements gathering is as much about filtering as collecting. Knowing when a request is a genuine business need versus a “nice to have” someone picked up from a competitor’s system takes judgment built from having been wrong about that distinction before. Gap analysis works the same way, the question isn’t just whether Business Central can do something out of the box, it’s whether building a customization is worth the maintenance cost it will create for the next decade.

Solution design means balancing customization against long-term maintainability, which is rarely a clean trade-off. Data migration planning means anticipating the legacy data quality problems nobody mentions in a requirements document because they’ve stopped noticing them. Integration strategy, security design, user training, go-live planning, hypercare, none of these are solved by generating more code. They’re solved by understanding the business well enough to make calls that have no textbook answer, and by being willing to have uncomfortable conversations when a stakeholder’s request doesn’t hold up.

Where AI Creates Tremendous Value

None of this makes me an AI skeptic. Quite the opposite, AI has become a genuine accelerator in Business Central implementations, and any consultant ignoring that is making their own life harder than it needs to be.

On the technical side, AI is excellent at generating AL extensions for custom fields, scaffolding permission sets, and drafting report layouts that used to take a developer half a day to get right. It’s useful for producing first-pass API integration examples, and for building out the initial logic in a Power Automate flow before a consultant tunes it to the client’s actual approval hierarchy. Functional specifications, test cases, code reviews, performance optimization, AI speeds up all of it, often meaningfully.

A practical example: instead of a developer spending hours writing repetitive logic for a custom approval workflow, AI can draft the base structure in minutes. The consultant then spends their time refining the business rules that actually matter, who approves what, at what dollar threshold, with what exceptions, rather than fighting with syntax. That’s a real shift in where the hours go, and it’s a good one. It means more time on business process optimization and less on the mechanical parts of getting there.

Why AI Cannot Replace ERP Consultants

Here’s where the limits show up, and they’re not subtle once you’ve sat through a few of these conversations yourself.

AI doesn’t have business judgment in the sense of weighing genuinely competing priorities, inventory accuracy against faster order fulfillment, say, when improving one makes the other harder in the short term. It can’t manage stakeholders who don’t trust each other, and it can’t deliver bad news to a board in language that keeps their confidence intact. Change management, negotiation, reading the room when someone’s nodding along but clearly isn’t on board, these aren’t technical problems, and treating them as if AI could solve them misunderstands what actually derails ERP projects.

I remember one implementation where finance wanted monthly closing tightened up, and the warehouse team was quietly dreading it, they were worried that mandatory physical stock counts would grind their operations to a halt during an already busy season. No AI tool resolves that. It took a series of conversations, a compromise on count frequency during peak periods, and someone in the room who both sides trusted enough to broker it. That’s leadership, not logic, and there’s rarely a single correct answer to a situation like that, just a workable one that both sides can live with.

The Business Value Great ERP Consultants Deliver

When these projects go well, the payoff shows up in numbers a CFO cares about, not in the codebase. Cash flow improves when receivables processes get tightened. Inventory drops when planning actually reflects real demand instead of habit. Procurement gets stronger when supplier performance is tracked instead of assumed. Operating costs come down as standardized processes replace the patchwork each department used to run on its own. Reporting gets faster and more trustworthy, adoption improves because people were brought into the process early instead of handed a finished system, and project risk goes down because someone experienced was watching for the pitfalls before they became problems.

None of that is abstract. It’s what happens when Microsoft’s ERP technology gets paired with someone who understands the business well enough to configure it around how that business actually works, not around a generic template.

The Future ERP Consultant

I don’t think the consultant’s job is disappearing, I think it’s shifting toward something better. The next generation of Business Central consultants will lean heavily on Microsoft Copilot for configuration work, use AI agents to keep an eye on transaction patterns, build low-code automation on the Power Platform, and use Power BI to make the business case for change with real data instead of just asserting it.

That’s a good direction. It means less time spent on the mechanical, repeatable parts of implementation and more time spent doing the part of the job that actually requires a person, advising, prioritizing, and helping an organization figure out what it actually needs, not just what it asked for.

Case Study: A Distribution Company Implementation

A mid-sized distribution company came to us with a familiar list of complaints: too much manual order entry, poor visibility into inventory across locations, and a month-end close that dragged on far longer than it should. During discovery, it became clear the real issue was fragmentation, sales, purchasing, and inventory were each running off their own spreadsheets, reconciled by hand, usually under deadline pressure.

We recommended implementing Business Central with standardized workflows across those three functions, rather than trying to preserve each department’s separate process inside the new system. AI played a real role in the build: generating the initial AL extensions for a custom pricing rules engine and drafting the first version of the Power Automate flows for order approvals, both of which our developers then refined against the client’s actual pricing logic and approval hierarchy.

User acceptance testing was led by consultants who made sure the test scenarios reflected how staff actually worked, not just how the process was documented on paper, which is where a fair number of gaps tend to surface. The system went live with fewer disruptions than the client expected, and hypercare support over the following weeks focused less on fixing bugs and more on coaching people through the parts of the new process that still felt unfamiliar.

Within a few months, order entry time had dropped substantially, inventory accuracy improved to a level the operations team was finally comfortable relying on, and month-end close shortened from roughly ten days to somewhere in the four-to-five day range depending on the month. None of that was really about the software. It was transformation, guided by people who understood the business, and accelerated, but not replaced, by AI.

Common ERP Project Mistakes

One lesson many ERP teams eventually learn, usually the hard way, is that most of the damage happens before a single line of code gets written. Treating ERP as an IT project rather than a business one sets the wrong tone from day one. Excessive customization creates fragile systems that are painful and expensive to upgrade later. Poor master data quietly undermines every report the system produces, long after anyone remembers where the bad data came from. Weak executive sponsorship leaves a project without the authority it needs when priorities inevitably clash. Inadequate training leads to low adoption no matter how well the system was built, and ignoring change management is often the single biggest reason a technically sound implementation still fails to deliver value. Choosing the software before understanding the processes it’s meant to support is the mistake that tends to cause all the others.

A common mistake we see is assuming that avoiding these pitfalls is mostly a project management exercise. It isn’t. It takes someone who has been through enough of these implementations to recognize the warning signs before they become expensive.

Conclusion

AI will not replace ERP consultants. It will make good ones considerably more effective, and it will expose the ones who were never doing much beyond writing code in the first place. The consultants who thrive from here will be the ones who pair genuine business expertise with AI-powered delivery, using the technology to move faster without losing sight of why the project exists.

The greatest value in ERP consulting has always been understanding businesses, improving how they operate, managing the human side of change, and helping people succeed with a new way of working. That was true before AI, and in my experience, it’s more true now, not less.

I’d welcome hearing how others are navigating this shift, whether you’ve seen AI change the shape of your own ERP projects, or whether the harder parts of implementation still feel exactly as human as they always did. If you’re weighing a Microsoft Dynamics 365 Business Central implementation or an ERP modernization effort, I’m glad to compare notes.