There is a running joke that the most popular ERP system in the world is not SAP or Oracle. It’s Excel. Walk almost any plant floor and you will find spreadsheets doing real work: tracking consigned stock, logging quality inspections, managing supplier approvals. Those spreadsheets exist for a reason. They were built by the people who understand the work, at the speed the work demanded. 

That last sentence is worth sitting with, because it explains the most exciting shift happening in business software right now. 

The game of telephone is ending 

For forty years, business software has been built to spec. Business people describe what they need, requirements get written down, handed to analysts, then to developers, then tested and delivered. Every handoff loses a little meaning. By the time the software ships, months have passed, the budget has doubled, and the business has already moved on to new problems. 

There’s a line we keep coming back to:  

Developers build what the documentation defines. Business process owners build what the work actually demands. 

The spec is an approximation, a translation of a translation. The process is the real thing. Every spreadsheet on every plant floor is proof of what happens when the person who knows the process finally gets a tool they can shape themself. The tragedy is that until now, the only such tool was a spreadsheet. 

The builder is changing 

AI has changed who gets to build. New AI powered tools can turn a plain language description into working software. You don’t write code, and you don’t need to learn how. You describe what you need, look at what comes back, and refine it, which is often referred to nowadays as vibe coding. What it means for the everyday ERP user is simple and honestly a little thrilling: the person who owns a process can now oversee the building of the tools that support it. 

Notice the word oversee. The process owner does not become a programmer. She becomes something more valuable: the architect of her own workflow. She reviews what gets built the way she would review a draft from a colleague, catches what is wrong immediately because she lives the process every day, and sends it back with corrections in plain English. The feedback loop that used to take months now takes an afternoon. That speed does not just make building cheaper. It makes experimentation possible. You can try an idea, discover it is wrong, and try another version before lunch, which is how good tools have always been made. 

If you can visualize it and articulate it, you can build it. That is a sentence that has never been true in enterprise software until now. 

The art of the possible 

Think about what that opens up: 

  • A quality manager describes exactly how inspections flow on her floor, and gets an inspection workbench that matches it, connected to live inventory, with nonconformances routed automatically to the right approver.
  • A planner starts each morning with a screen built around his role, showing precisely the items that need action today, the orders approaching risk, and the purchase triggers that fired overnight, instead of a generic report he has to dig through.
  • A supply chain lead finally retires the consignment spreadsheet, replaced by a tracker that handles customer owned stock, consumption, and billing in one place.
  • A finance team connects their inbox to the ERP so that vendor invoices automatically route to the right approver and then land in the three-way matching workbench.

None of these have to be features waiting on a vendor roadmap anymore. They can now be conversations between a process owner and AI, turned into working software in days. 

And that list is only the obvious part. The bigger shift is what happens to everything that never made it that far, the ideas too small, too specific, or too particular to one business to ever earn a spot on somebody else’s roadmap. That pile is enormous, and it is where the real opportunity lives, because it’s also where your special sauce lives: the steps and workarounds that are truly unique to your business. ERPs have always forced a choice here, bend the process to fit the system or bend the system until it breaks with the next upgrade. That choice is gone. 

And because that choice is gone, day one, day 365, and year five do not have to look the same. What you build to match how the business runs today will not become tomorrow’s constraint. Add a product line, enter a new market, find a better way to run a process, and the system can easily grow with you instead of making you wait for it. 

Take away the cost and the wait that used to decide what got built, and the only thing left deciding it is your imagination. 

Why not just build the whole thing yourself? 

Some ambitious teams are already asking the obvious next question: if AI can build applications, why not vibe code our own ERP from scratch? It is a fair question, and the honest answer is that the economics do not work. An ERP is not one application. It is decades of accumulated logic about costing, planning, compliance, traceability, and thousands of other things that are invisible until they are wrong. Building that from nothing means carrying all of the development cost, all of the AI usage cost, and all of the risk, alone.  

The smarter math says to start from a foundation that already handles what every manufacturer needs, one that was built to be extensible from the ground up, and pour your energy into building the twenty percent that makes your business uniquely yours. 

Why the foundation underneath matters 

This kind of AI-enabled product extensibility is only as trustworthy as the ground it stands on. If it gets built outside your core system, you have just reinvented the spreadsheet problem with better graphics: separate logins, another copy of the data, and no oversight. If it is built inside a platform that shares one security model, one set of user permissions, one source of data, and one governance process, then everything you create behaves like part of the system, because it is part of the system. The same person sees the same data under the same rules, whether the screen came from the vendor or from your own team. And your IT and admin teams stay in the loop, reviewing and deploying what gets built through the same controls they already use, which turns them from a bottleneck into an enabler. 

This is where the market quietly splits. Most ERP systems treat extensions as outsiders that connect from a distance. A smaller group of systems live natively on platforms that were designed from the start for safe extension, where anything you build automatically respects the same users, roles, approvals, and admin controls as the ERP itself.  

Rootstock sits in that second group because it is built natively on the Salesforce platform, so everything customers create inherits the same security and governance as the ERP they already run. This trend rewards foundations that were built for it, and no amount of AI bolted onto a closed system changes what the system is underneath. 

Who maintains it when the builder moves on? 

It is the right question to ask, because the reason customization burned a generation of manufacturers wasn’t just the build. It was the upkeep. Thankfully, extensions built on a governed platform age differently. They are visible, documented, and administered through the same controls as the ERP itself, which makes them company assets rather than secrets on someone’s hard drive. And the same AI that helped build an application can read it, explain it, and modify it, so the next process owner can pick up exactly where the last one left off.  

A more interesting question 

The ERP conversations of the next few years will not be about module checklists. They will be about imagination. The gap between I wish the system did this and the system does this is shrinking from years to days, and the people closing it are not developers. They are the people who know the work best. For anyone who has spent a career maintaining workarounds, that is genuinely groundbreaking, and it’s only the beginning. 

If the word customization still makes you nervous, we understand. That instinct made sense for decades. We wrote about why it no longer does: read it now, Why Customization Is No Longer a Dirty Word.