If you're investing in a system to manage your business operations (an ERP system) right now, it's a fair question: will this be money wasted in two years, once AI takes over everything anyway? A lot of Hungarian business owners delay the decision for exactly this reason. My own experience shows something different. AI won't replace your ERP: it builds on top of it and changes how you work with it. The real question isn't 'ERP or AI'. It's whether your data is in order: if it isn't, AI won't work miracles either. In this article, I'll share what I'm seeing in the market and in the day-to-day operations of small and medium-sized businesses (SMEs) in Hungary, and what it means for your company.
The question on a lot of business owners' minds
The pace of AI development is both frightening and exciting. If you're about to roll out a major system, it's fair to pause for a moment and ask: is it still worth doing, if AI is going to 'change everything' tomorrow anyway?
This isn't a theoretical worry. Plenty of Hungarian SMEs postpone real decisions because of it: they'd rather wait than bet on the wrong horse.
In my day-to-day work, this is the uncertainty I see most often. So it's worth saying plainly what I think, based on what I see in the market and the systems I've helped businesses put in place: AI isn't making ERP obsolete. It's changing how you work with it.
The system isn't dying, it's transforming
The idea that 'everything will be rebuilt on the ruins of the old system' may sound compelling, but it's an exaggeration. What I actually see in the market is the opposite: ERP has a future, and demand for AI solutions built around it keeps growing.
What is genuinely changing is the division of labour. It's increasingly likely that certain tasks won't be handled by the ERP itself, but by a connected AI assistant. Today, logistics, compliance with rules and regulations, and security are often handled separately: AI can link these areas together. The system stays, but its boundaries shift.
2026 to 2027: when assistants enter the system
Over the next two years, many companies will meet real AI assistants inside their business systems for the first time. The major AI developers have released the first assistants that don't just answer questions but also carry out tasks for you inside the system: filling in a form or starting a process, for example.
Larger companies are already testing several such AI assistants for different processes. That sounds intimidating at first, but for an SME it's actually good news: you don't need a dozen assistants. A single, well-chosen process, say processing incoming invoices or preparing quotes, is enough to get your first tangible benefit.
The lesson isn't 'rush'. It's to prepare for the next few years with an organised system and organised data.
The real obstacle isn't AI, it's data
Here's the part people rarely say out loud: most companies, especially smaller ones, can barely make use of their own data. It's scattered, unorganised, sitting in half a dozen different places.
And this is exactly where AI fails most often. It isn't a magic wand: how useful it is depends on how well organised the data it works with is. If the order sits in one spreadsheet, the invoice is with the accountant, and the customer correspondence is in an inbox somewhere, even the most advanced AI won't know what to do with any of it.
That's why I think the order matters and can't be reversed. First, you need one shared system where you see the order, the invoice, the customer and the stock in one place, and only then is it worth building AI on top. For me, that shared system is Odoo, not because it's 'new', but because it puts everything that's scattered today into one place. AI comes after that, wherever it actually delivers value.
Replacing your old system without going through the same pain again
Many business owners don't dare replace their old, inherited systems because the memory of a previous rollout still stings: a drawn-out, expensive, risky project that ended with a solution that only partly met the company's needs.
The good news is that this part is changing too. Building on the knowledge accumulated in your old system, AI can now help you redesign your processes in several different ways. You can test which version actually works, without committing to a multi-year, brutally expensive project.
For me, this isn't new: it's how I work. I roll out systems in small steps, breaking the work into milestones, so there's a tangible result early on instead of a surprise two years down the line. In one earlier project, this approach got a company to a working solution within a few months, whereas previously, the same implementation would have taken years. The secret isn't the technology, it's how you move forward.
Who owns your data? The dependency you should factor in now
There's also an uncomfortable but real question: who actually owns the data stored in the system, and who gets to manage it? With several vendors, this is becoming an increasingly pressing issue: the provider can even restrict how much access a customer has to their own cloud-stored data.
There's a financial side to this too. If a company sends, say, all of its invoices through its own digital assistant to a closed, proprietary external service, that can tie up a surprising amount of capacity and come with significant cost. Convenience has a price, it just doesn't always show up where and when you'd expect.
This is exactly the question that I explore in a separate article : when it's worth running the critical, sensitive parts on your own systems, under your own control, instead of a closed, proprietary cloud environment run by an external provider. If you're rolling out a system now, it's worth thinking through from the start where your data will live and how long you'll be able to access it.
A new role that's already taking shape
Further down the road, digital assistants will become more autonomous and start improving processes on their own. The person running the company won't necessarily be a narrow specialist, but someone who understands how work flows through the business: someone who sees where the work gets stuck and where unnecessary steps can be cut.
This is a message for SMEs too. You don't need to be a futurist right now, but it's worth organising your data and shaping your processes today so there's something to build on tomorrow. Getting your house in order is worthwhile in itself. It also gives AI a home it can move into.
What does this mean for your company?
Here's the practical version. AI won't replace your ERP system: it builds on top of it and changes how you work with it. The system stays the backbone of your operation, and AI is the layer on top that speeds up the work and takes the repetitive tasks off your plate.
There's one thing I'd warn you about: if you don't organise your data now, you'll fall behind. Not because AI takes your job, but because your competitor will get to work with organised data sooner.
So here's the order I recommend. First, get your data in order: in one system, in one place. Then bring in AI wherever it delivers real value: invoice processing, quote preparation, customer service. And do it all in small steps, so there are results along the way, not just at the end. That's what I offer at Glazer-Innovation Kft.: not AI instead of ERP, but ERP and AI together, tailored to Hungarian SMEs.
Frequently asked questions
Will AI replace ERP systems?
No. The system isn't disappearing. If anything, demand for AI solutions built around it keeps growing. What's changing is the structure: certain tasks get done elsewhere and differently, and more and more AI assistants work inside the system. The ERP system stays the backbone of operations.
If I'm rolling out an ERP now, should I wait because of AI?
It's actually the opposite. AI works with organised data, so you need the system and organised data first, and AI can build on top of that later. Waiting isn't an advantage: an organised setup is exactly what lets your competitor pull ahead.
How urgent is this for an SME?
Over the next few years, AI assistants will show up in more and more business systems. There's no need to rush, but getting your own data in order is the step worth taking today, so there's something to build on next year.
Isn't it risky to hand company data over to AI?
That's a fair concern. Where the AI runs and where the data ends up matters a lot: it can be worth keeping the sensitive part in your own, controlled environment rather than a cloud environment controlled by an external provider. I've written about this separately too.
If you want ERP and AI: from a single source
At most companies, ERP and AI are handled separately: the system comes from one vendor, the AI from another, and in the end no one takes responsibility for the solution as a whole. I bring these together: one shared system (Odoo) with AI built on top, from a single source, in Hungarian, tailored to your company.
Request a free consultation: in one conversation, I'll look at how you currently operate and tell you where using AI will deliver the quickest return.
If you'd rather see it in action first, here's what a business AI assistant looks like in practice in my work. The consultation is free and comes with no obligation.