Are you a digital leader about to tell your CEO, for yet another quarter, that your AI initiatives haven't delivered the transformational growth you promised? Then this post is for you.
By the end, you'll understand what it actually takes for AI initiatives to drive growth. You'll have a new way of thinking about AI use cases and features, and a specific, actionable plan to make it happen.
This isn't analyst fluff. It's what we see in the trenches, working with teams that ship AI features every day.
This is the first post in a series that aims to give digital leaders the straight answers on enterprise AI they aren't getting elsewhere.
The promise vs. the reality
It's been almost four years since large language models (LLMs) broke into the mainstream. AI-enabled applications showed the world the revolutionary potential of generative AI, and the markets took note. Enterprises steadily invested in AI to drive operational efficiency and transformational growth. Technology vendors added AI features to their products.
Despite the promises, the financial impact of AI has not been revolutionary:
- Research shows that AI initiatives are not contributing to EBITDA for two out of three enterprises.
- The operational costs of using AI have been rising.
- 20% of enterprises are looking to scale down their AI use, even though they are expected to increase their AI investments in the years ahead.
Yet some organisations have deployed AI with great success. Research shows that these high performers differ in how, and to what extent, they deploy AI. They combine efficiency with growth, and they redesign existing workflows so that AI enables them. The European Personnel Selection Office (EPSO), which we cover later in this post, is one example.
This post explains why many enterprises struggle to use AI to become more efficient and grow the bottom line. It also introduces an operational model for moving from individual productivity to financial impact.
Where conventional wisdom gets it wrong
AI is not stardust you sprinkle on top of your organisation to magically make things better. Nor is it about adopting new technology as fast as you can. AI is a transformational technology, with potential for both growth and disruption.
Integrating AI into an enterprise means rethinking which tools, processes and skills you need. It's a change management process as much as a technology adoption journey. Luckily, it's not the kind that requires very expensive consulting work. It does require both a business and a technical lens.
That's what most enterprises get wrong. Enabling an AI assistant in your CRM or generating blog posts with Claude might sound cool. But it may bring hidden challenges that stop the enterprise from growing. Here are some of them.
1. People: AI literacy
Not everyone sees AI as an opportunity. Across an enterprise, many workers see it as a threat. That makes them less likely to adopt it, become proficient with it, or bring it into their own processes.
That's why AI literacy is vital. People need to understand how to use AI, when to use it, and how it can make them productive.
Digital leaders can turn this weakness into an opportunity. Plan for training. Use AI to solve some of the most annoying operational problems in the enterprise. People will thank you for it, and you'll break down barriers much faster.
2. Data: knowing what you have
For AI to be specific to your organisation, it needs your data, context and processes. If an enterprise doesn't know which data it produces, who is responsible for it, and how it can be used across departments, all it will manage is access to the same generic AI tools as everyone else. That means no competitive advantage.
Data used to live in databases. With AI, videos, internal knowledge bases and digital documents are all data too: multimodal data, to be exact.
To use AI effectively, enterprises need an inventory of their data and a mechanism to share it. That makes data literacy vital as well, along with some investment in a data programme that monitors how data is created and used.
3. Architecture: avoiding a disconnected landscape
There's a lot of hype around AI. Out of fear of missing out, many enterprises rush to enable it in their existing tools. That's fine as a first step. In fact, it's a very good way to familiarise the enterprise and its people with the power of the technology.
But when every disconnected tool in the organisation has some AI, you end up with a disconnected landscape where it's very difficult to create value. At most, you get some minor efficiencies.
Building sturdy AI foundations means taking a step back and examining your digital landscape from a technical architecture perspective:
- Do you have systems of record?
- Are they integrated?
- Can your customer-facing systems source data from back-office or legacy systems?
If the answer is no, the issues you face today will carry over into your AI.
4. Strategy: a shared view of where value lies
Does your enterprise share an understanding of where AI could create the most value or bring the most efficiency? A clear goal with a simple plan outperforms a grand scheme that is hard to execute, every time.
This is where zero cost thinking comes in handy. Ask yourself: if the cost of any development were zero, which project would create the biggest value for the enterprise? AI tools are reducing the time and cost of technical implementations, so more value can be achieved with the same timeline and budget.
A new model for AI adoption
Zero cost thinking helps companies explore potential use cases from a strategic perspective. But a few first steps are needed to make sure the right foundations are in place, so that initiatives can deliver value.
At Dropsolid AI, we approach AI adoption as a series of steps. Some are quick wins that are easy to fix. Others require larger projects. Our expertise is in identifying the quick wins and helping companies prioritise everything else.
To see where you stand, it helps to understand the three phases of enterprise AI adoption.
Phase 1: Get to know AI
Most enterprises are here. They have started deploying AI, mostly inside the digital tools they already use, and are building excitement about its capabilities and opportunities. This phase brings small but measurable results, with productivity increases reported at 10% to 15%. But results like these don't register on EBITDA. That's where the next phase comes in.
Phase 2: Transform core functions
Most value would come from transforming core business functions. Enterprises further along in their AI adoption use AI to combine efficiency with growth and to redesign existing workflows. (The EPSO case below shows what this looks like in practice.)
Phase 3: Prioritise use cases
With core functions transformed, an enterprise can invent new offers, services and experiences with AI, and capitalise on a new era of innovation to significantly transform the business.
Every company's journey is different. It depends on their digital platforms, how well their systems are integrated, and which AI technologies are available in their technical stack.
Find out where you stand: the AI readiness benchmark
To capture those differences, we run a short AI readiness benchmark. It takes about 30 to 45 minutes, it's guided, and I run it myself. My background is a PhD in machine learning, so it's a genuinely technical read rather than a sales pitch.
You come away with a clear picture of your data, your team's AI fluency, and the use cases closest to real value. That gives you real metrics instead of a hunch, and something concrete to take back internally.
In practice: how EPSO transformed two core functions
So what does a use case that transforms a core function look like?
The European Personnel Selection Office (EPSO) is the EU body responsible for recruiting staff for all EU institutions. During recruitment phases, hundreds of thousands of people visit its website and contact the support centre with questions.
EPSO wanted to use AI to transform two core functions:
- Faster, more accurate answers from support officers: Waiting days for an answer to a question that should take minutes erodes confidence in the process. For an EU institution operating in 24 languages, response time and clarity matter for the integrity of the selection process itself.
- Self-service for website visitors: A high proportion of incoming questions, especially routine questions about procedures, can be answered from EPSO's published procedural documents. But candidates have to find those documents, navigate to the right section, and understand the content in plain language.
To do this, EPSO partnered with Dropsolid AI and deployed two AI solutions across 24 EU languages, on EU-sovereign infrastructure:
- Knowledge Navigator: Staff handling candidate questions can ask the assistant in plain language. They get an answer grounded in EPSO's own procedural documents, with citations to the relevant section.
- Trusted Answers: Candidates ask procedural questions in their own language. They get answers drawn only from EPSO's published documentation, with citations they can click through to verify.
The result: questions reaching support dropped by 70%. Support staff can now spend 100% of their time on the remaining 30%, the more complex questions. It's a clear case of combining efficiency with growth, where savings on one side are invested on the other.
What this means for EU enterprises today
Industry reports show high-performing enterprises achieving transformational growth with AI. But European enterprises have different terrain to navigate than their North American counterparts.
Most AI services are part of public cloud hyperscalers, or come from suppliers established or controlled in third countries. Some of these may raise cybersecurity concerns. EU companies should assess sovereignty, control and supply-chain transparency against specific assurance criteria.
In the long term, that may not be as complicated as it sounds. But it is a substantial task at the start of an AI journey.
Conclusion
Enterprises that don't see the results they hoped for from AI fall short because of their approach. Adopting AI is a journey, not a destination. Depending on its maturity, an organisation will need to stop at different phases for different amounts of time.
It takes people with the right knowledge and skills, and leaders with a clear map of where they begin and where they will end up.
Ready to move from efficiency gains to growth?
If any of this sparked your interest, drop me a line. I'm happy to tell you more about the AI readiness benchmark: about 30 to 45 minutes, guided, giving you a clear picture of your data, your team's AI fluency, and the use cases closest to real value.