Blog · November 4, 2026
In our last article, we argued that AI is turning the ERP into a professional's tool: instead of learning dozens of screens, people can ask their systems questions in plain language.
A natural follow-up question is: if AI can work with an ERP so easily, can it also implement one? Will implementations become a fraction of the time and cost?
Our honest answer, as an implementation partner: AI makes part of the work much faster. The part that decides whether an ERP succeeds stays very much the same.
The build is the smaller part of the project
From the outside, an ERP implementation looks like configuration: setting up modules, building forms, writing workflows and custom code. That is the part AI speeds up most. AI assistants connected to the platform can now set up structures, generate test data and write code in a fraction of the time it used to take.
But in our experience, configuring, customizing and coding is 20 to 30 percent of the effort in an ERP implementation, at most. The rest goes into:
- requirement sessions with each business function
- solution design, and the back and forth that refines it
- writing and reviewing documents
- conference room pilots (CRPs) and user acceptance testing (UAT)
- progress meetings, internal reviews and decisions
- data migration, training and go-live support
The arithmetic is simple. If the build is a quarter of the project and AI halves it, the project gets about 12 percent shorter, not 50 percent. That is a real gain, and clients benefit from it. But anyone promising that AI will collapse an ERP timeline is talking about the smallest part of the work.
The true value is in advisory
The time in an implementation goes into people, decisions and validation, and that is where the value is too.
A good partner brings the experience of many implementations: what works in your industry, how other organizations have solved the same problem, which shortcuts come back to hurt you, and which requests are worth pushing back on. They bring an approach: a way of moving an organization from its current process to a better one without breaking what already works. And they guide you through the decisions that only you can make.
None of that is generated. It comes from having done it before, many times. As AI makes the build cheaper, this advisory part becomes a larger share of what you are really paying for, and the most important thing to judge when choosing a partner.
Requirements are still gathered by listening
There is a popular idea that AI-built prototypes will replace requirement workshops. For ERP, we do not agree.
Every business function is a little different. Finance, procurement, HR and sales each have their own logic, exceptions and history. Capturing that has always been a consulting skill: experienced consultants listening, asking the right questions, logging it carefully, and using a short demo where it helps people picture the outcome.
Prototypes have their place in custom development. In an ERP project they mostly add effort. Real feedback only comes when people use the complete system with their own data, which is exactly what CRPs and UAT are for. A partial prototype in between is a tool, not a substitute for that.
The system still needs its structure
If AI can answer questions about your business, why not just give it the data and skip the design?
Because a business system cannot be a blank file structure. For AI, or people, to use an ERP reliably, the processes, data structures, forms, approvals and controls have to be designed in, aligned with your stakeholders' needs and with best practice. That structure is what makes an AI answer trustworthy. Without it, you get fast answers built on inconsistent ground.
Designing that structure is the heart of an implementation, and it still requires people who understand both the platform and your business.
Training is still training
Good ERP training was never only about which button to press. It covers the tool together with the process, the policy, the best practice, the do's and don'ts, and what to expect when something goes wrong. User manuals and role-based training remain as important as ever.
What we add now is an introduction to AI access: showing users that they can ask the system questions, and encouraging them to be curious. It sits alongside the training, not in place of it.
What does change
To be fair to AI, some things do improve:
- A faster, more consistent build. Configuration and code are done faster, with fewer manual errors.
- Better use after go-live. Once the system is in place, AI lets people get far more out of it, as we described in our previous article.
What to look for in a partner
When you choose an ERP partner in the AI era, ask about their AI capabilities, by all means. But ask the old questions too:
- How do you gather requirements, and who will be in the room?
- What is your approach to CRPs, UAT and change management?
- What experience do you have in our industry and region?
- Who owns the solution design, and how are decisions recorded?
AI is changing how ERPs are built and used. It has not changed what makes an implementation succeed: experience, advice and people who take responsibility for the outcome.
How we can help
IT Enablers Global is a Zoho Premium Partner. We implement Zoho as a complete ERP.
If you are planning an ERP implementation and want an honest view of where AI helps and where it does not, talk to us.