What Your AI Maturity Score is Actually Telling You

Ask a room full of senior leaders to rate their organization’s AI maturity on a scale of one to five, and most will fall between a two and a three. Ask the same group to rate their ERP maturity, and the numbers shift — a three or four is more common. That one-point gap may not sound like much, but it appears consistently enough across different industries and company sizes that it’s worth asking what it’s actually indicating. We think it’s telling you something worth paying attention to.

What the Five Stages of AI Maturity Look Like

Here’s what each maturity stage looks like on the ground.

A one is exactly what it sounds like — no formal AI strategy, most processes running the way they’ve run for years. A lot of well-run companies are here, particularly those that have been heads-down on ERP consolidation or working through a major acquisition cycle.

A two has a texture most leadership teams will recognize, even if they haven’t named it. Employees are using AI tools like Copilot, ChatGPT, or something the marketing team found and expensed without telling IT. Still, there’s no governance, no shared framework, and limited visibility into what’s actually happening across the organization. Shadow AI is almost always more widespread than leadership realizes.

A three is where things get intentional but not always unstuck. Pilots are running. There’s probably a cross-functional committee, or at least a conversation about forming one. Some of those pilots are genuinely working, and somehow still haven’t scaled six months later. This is where many capable organizations are quietly frustrated.

A four is where the ERP investment starts paying its AI dividend. Solutions built for one business unit extend to others because the underlying data infrastructure actually supports it. AI is moving across the organization, not just through pockets of it.

A five refers to a state where AI is embedded in core operations, aligned with strategy, and championed from the top, at least in theory. In reality, when we ask rooms full of senior leaders if anyone’s there, the honest answer is almost always no. It’s a useful guiding principle, but not a realistic near-term goal for most.

What the Gap Between Your Two Scores Is Telling You: ERP vs. AI Readiness

If your ERP maturity is a three or four and your AI adoption is a two or three, the gap usually comes down to one of two things, and it’s worth figuring out which.

The first possibility is that the foundation isn’t as solid as the ERP score suggests. A company can be running a modern ERP and still have data quality problems, unconsolidated business units, or historical records that never got properly migrated. The system is there, but the data inside it isn’t ready. In that case, the AI score isn’t lagging behind the ERP score, it’s accurately reflecting the real state of the foundation beneath it.

The second possibility is that the foundation is genuinely solid, but the organization hasn’t figured out how to move from infrastructure to action. The data is clean, the systems are consolidated, but there’s no clear framework for identifying where AI creates real business value versus where it just looks impressive in a demo. This is a strategy and prioritization problem, not a technology problem, and it has a different solution.

Knowing which situation you’re in changes what you do next. If it’s the first, more AI tools won’t help — the work is in the data. If it’s the second, the foundation is there and the opportunity is real; what’s missing is a structured way to find and sequence the right use cases.

Why So Many Organizations Are Stuck in Pilot Purgatory

Most “threes” have run AI pilots. Some of them worked remarkably well in their original context. The problem is that “working well in a pilot” and “being ready to scale” are not the same thing, and the distance between them tends to be larger than expected.

Pilots succeed in controlled conditions, such as a single business unit, a clean dataset, a team that was selected partly because they were enthusiastic. Scaling requires something different: consistent data across every entity the tool needs to touch, organizational alignment that goes beyond the pilot team, and a clear definition of what success looks like at scale before you start building toward it. Most pilots don’t define that last part upfront, which means there’s no clear trigger for when a pilot becomes a program. It just… continues. Gets refined, generates positive feedback, but never quite graduates.

Why Successful AI Pilots Fail to Scale

This is pilot purgatory, and it’s where a lot of genuinely capable organizations are sitting right now — not because they’ve done anything wrong, but because the path from a successful pilot to scaled deployment isn’t as obvious as it looks from the outside.

A few questions often reveal where the real bottleneck is:

• When your last pilot succeeded, what would have had to be true for you to roll it out to two other business units within six months? Was that path clear at the time?
• Does your AI data sit in the same environment as your operational and financial data, or are they still separate?
• If you built something for one business unit today, how many integration points would you need to build before another unit could use it?
• Do you have a cross-functional group with real decision-making authority over AI initiatives, or is AI still happening in functional silos?
• When you evaluate an AI pilot, are you measuring against defined business outcomes, or against whether the tool works as advertised?

There are no right answers here. The value is in being honest about what the answers reveal. An organization that can answer those questions clearly, even if some of the answers are uncomfortable, is in a much better position than one that hasn’t asked them yet.

What Moving Forward Requires

Microsoft describes the path toward what they call a Frontier Organization in three phases — from using AI to retrieve and summarize information, to automating significant parts of business processes, to eventually letting AI run full workflows with humans in a supervisory role. It’s a useful framework not because it tells you anything surprising, but because it makes explicit what often stays fuzzy: these are genuinely different stages of maturity, and the work required to move between them is different each time. There’s no shortcut from phase one to phase three.

The companies we see making that progression tend to share a few habits. They define the business problem before they evaluate tools, not the other way around. They do an honest assessment of whether their data environment can actually support the use case before committing to it. They build enough governance to know what AI is being used across the organization, because the alternative is shadow AI proliferating in ways that create real data exposure. And they decide what “ready to scale” looks like before the pilot launches, so there’s a defined moment when a successful experiment earns the right to become something bigger.

A two or three right now puts you squarely in the middle of the market — that’s just where most organizations are. McKinsey’s 2025 State of AI research found only about 1% of companies describe themselves as mature in AI deployment, meaning AI is fully embedded and producing measurable business outcomes. That’s not a gap most organizations are going to close overnight, but it’s also not as wide as it feels.

The companies that make genuine progress over the next couple of years probably won’t be the ones with the biggest AI budgets. They’ll be the ones who got honest about their data foundation and dealt with it — a path which, more often than not, means addressing any gaps in their ERP strategy.

Velosio works with companies at every stage of the AI and ERP maturity journey. If you’d like an honest assessment of where you stand and what moving forward looks like, take our AI Maturity Assessment.