It Started With a Licence—and Ended With a Realisation
Not long ago, I sat in a boardroom with a leadership team who had just unveiled their proudest digital investment of the year: Co-pilot licences for every employee across their company. It was a bold move—hundreds of thousands of euros committed annually, a statement that the business was “future-ready,” surfing at the crest of the AI wave. The intention? To supercharge productivity, propel innovation, and give their people a competitive edge.
Rollout day had all the trappings of a modern digital transformation—excited all-hands meetings, rousing emails, even a handful of enthusiastic early adopters queueing to log in and experiment. Then the momentum fizzled.
A month in, usage metrics trickled back. Of the hundreds granted access, only a tiny minority had even attempted to use the tool. The vast majority hadn’t gone much further than opening the app once. The ROI—on paper—was abysmal. “We thought people would just get on board,” the CTO admitted, frustration etched across his face. “Isn’t everyone talking about AI these days?”
And that’s not an isolated case. Across industries, from retail to finance to professional services, I’m seeing a pattern: significant investments in AI licences, minimal adoption, and a creeping suspicion among leaders that the hype might be outpacing the substance.
The Shiny Hammer Syndrome: Why Ad Hoc AI Leaves Most People Behind
Let’s cut through the noise—AI today is the business world’s favourite shiny hammer. Vendors pitch it as “plug and play,” and so many organisations, chased by FOMO (and some very persuasive sales decks), simply kit out the team with licences and expect magical transformation.
There’s just one problem: giving everyone a sledgehammer doesn’t mean everyone builds a cathedral. It doesn’t even guarantee a shed.
Skill Gaps and the Law of Diminishing Returns
In reality, skill with AI—especially generative tools like ChatGPT or Co-pilot—isn’t distributed evenly. There are always a handful of enthusiasts: the curious marketer who plays with prompts, the innovation manager eager to automate a slide deck, the product lead who feeds a brainstorm into a language model for fresh angles.
But the rest? They’re busy, they’re uncertain, and they’re resistant to yet another tool. For many, the friction is simply too high. Writing a halfway decent prompt can feel like learning to code from scratch—only without the training, documentation, or support. Inevitably, results vary: one staff member gets stunning, project-ready proposals in minutes; another gets word salad or off-brand nonsense and gives up.
What Skill Disparity Really Costs You
What does this look like in practice?
- The sales rep inputting prospect notes; one gets a list of sharp follow-ups, another a paragraph of fluff.
- The HR manager summarising employee surveys; one extracts meaningful insights, another generates vague platitudes.
- The finance analyst querying expenditure; one crafts a prompt for actionable forecasting, another produces something even Excel shakes its head at.
Inconsistency like this isn’t just frustrating—it can actively sow distrust for the entire initiative, demoralising teams, and leading to “quiet quitting” of new tech.
The Psychology of Tool Fatigue
And let’s not forget the emotional cost: “tool fatigue.” Every month, it seems, there’s a new SaaS darling perched in your browser bar. Each promises to save time, but, ironically, the time sunk in signing up, experimenting, failing, and moving on is rarely recouped. Many staff revert to tried-and-true habits—Outlook, Word, old WhatsApp groups—where things at least feel predictable.
Licence Overload: The Sunk Cost Leaders Don’t Want to Discuss
Now for the uncomfortable truth—the license bill.
The companies I speak with are spending eye-watering sums on AI platforms. It’s not uncommon to see budgets swollen by 20%, 30% or more for generative tools, with justifications like “we can’t afford to be left behind.” But when I dig into the reports, usage rates often languish south of 20%.
That’s not transformation; that’s shelfware on a grand scale.
Let’s break down the real costs:
- Licences: Self-explanatory.
- Training (or lack thereof): All the money spent on monthly fees dwarfs the (nonexistent) investment in actually upskilling staff.
- Support and Security: Shadow IT explodes as employees hack together their own workarounds, increasing the burden on your technical and risk teams.
- Change Management: The more failed launches you have, the harder it becomes to get buy-in for genuinely valuable digital investments in future.
- Lost Productivity: Time wasted wrestling with tools is time not spent serving customers.
Recently, at Omnitas, I reviewed the numbers for a mid-sized firm: €300,000 annually on generative AI licences, barely 15% adoption after six months. That’s €2,000 per active user—per month—without counting the indirect costs above.
Ask yourself: If this were a new ERP or CRM, would you accept that kind of performance for that price?
AI That Just Works: The Case for Integration Over Experimentation
So what’s the alternative?
It’s less about doing “more AI” and more about embedding the right AI in the right places so it simply powers up daily work—no extra clicking, no special prompting, no long-winded tutorials.
A Real-World Integration: Project Documentation Reimagined
Enter workflow-integrated AI. One of my favourite recent examples involved a client who, like so many others, had invested heavily in AI but seen minimal return. I proposed a different route: integrate OpenAI directly into their monday.com project management suite using Make as the middleware.
Here’s what changed:
Instead of asking every project manager to learn prompting, the process was automated.
Staff filled in a project brief as always—no new apps.
The integration triggered AI to pull in requirements, milestones, stakeholder info, and risks, generating a first-draft project document, every time, in company-standard language and format.
Team leads reviewed and fine-tuned, but the bulk of manual work was gone. Reports were more consistent, more compliant, and much, much faster.
This wasn’t a marginal gain. The team estimated saving 2–3 hours per project, multiplied across dozens of active projects monthly. Staff with zero AI background received the benefit, onboarding time decreased, and templates became ironclad. The integration could be refined and scaled up as new needs emerged—no need for every employee to become a prompt engineer.
The Power of Invisible AI
This is what I mean when I say “AI in your business DNA.” The ideal scenario is one where staff barely think about the AI—they just notice their work is easier, faster, and more reliable. The interface is unchanged; behind the scenes, automation does the heavy lifting. This is how you generate consistent, defensible, and scalable value.
Compliance and Control: The Hidden Battle With Fragmented AI Use
The “everyone prompts” philosophy, especially when paired with a wild assortment of AI apps, gives compliance teams sleepless nights—and for good reason.
What Happens When AI Isn’t Controlled?
- Sensitive client or HR data finds its way into unsecured, third-party tools.
- Outputs vary so widely that the risk of off-brand or even legally questionable communication skyrockets.
- Leaders suddenly can’t explain, standardise, or audit what’s going out the door—because everyone is doing their own thing.
Contrast that with integrated AI:
- Prompts, workflows, and data flows are centrally defined, reviewed, and improved.
- Access is restricted as needed; logs and versioning are maintained.
- Outputs can be QA’d, ensuring they match regulatory and company standards.
Imagining a Compliance Nightmare
Imagine an insurance firm where individual agents are using their own AI tools to draft customer policy summaries. Data privacy is suddenly at risk, and if the regulator comes knocking, explaining what’s happened may be impossible. Brand risk, legal risk, and operational risk multiply. Even if none of it goes wrong, the burden to prove compliance is far greater.
Integrated AI: A Risk Manager’s Dream
With workflow-centric AI, all usage is traceable and standards-controlled. There’s one source of truth, prompt logic that’s versioned and tested, and risk is isolated to the platform, not scattered throughout the business.
Why True Empowerment Isn’t Found in “DIY AI”
It’s very tempting, especially for leaders who want to encourage innovation, to simply hand out tools and cheer for “empowerment.” And yes, the tinkerers—the digital natives, the curious few—will dive in. But for lasting value, you need a framework.
The Tinkerer’s Paradox
I’ve seen firsthand how a few creative staffers push the boundaries. That energy is invaluable—it’s how some of the best bottom-up automation starts, both at Omnitas and with our clients. But left unchecked, every success story spawns five “reinvention the wheels,” with brilliant processes stuck in siloed spreadsheets, or prompts that only work for their creator.
The Path Forward: Capture, Scale, and Standardise
The answer isn’t top-down lock-in or ignoring the early adopters. It’s recognising promising experiments, capturing them, and integrating the best into company-wide workflows.
At Omnitas, we often run “AI champion” programs—identifying those with the knack and giving them a sandbox. The gold that emerges is then formalised, embedded into CRM or project management processes, and rolled out so everyone benefits. That’s how you turn sporadic innovation into permanent improvement.
Assessing Your AI Maturity: Is It In Your DNA?
Let’s get practical: how do you know if your business is actually reaping the benefits of integrated AI, or just adding to digital noise? It helps to look at your AI maturity across a few axes:
- Adoption: Are most people actually benefitting, or is it a niche club?
- Consistency: Are results predictable and on-brand, or a mixed bag?
- Compliance: Can you trace, audit, and explain your AIs’ role in key processes?
- Empowerment: Are new users able to get results from day one, or do they have to reinvent the wheel every time?
- Iterative Improvement: Do your best AI-powered processes get better as you learn, or are experiments abandoned as soon as someone leaves?
The journey goes something like this:
- Stage 1: Individuals experiment. Lots of curiosity, little consistency.
- Stage 2: Teams adopt, silos multiply. A few quick wins, but risk and redundancy increase.
- Stage 3: Organisation learns to recognise and capture strong use cases.
- Stage 4: Integrated, invisible AI—embedded in workflow, everyone benefits, new hires receive value on day one, risk and quality are managed centrally.
Where are you?
From a Chore to a Competitive Advantage
It’s a simple formula, but a powerful one:
The companies that win with AI are those that make it reliable, invisible, and an organic part of business as usual—not a flag planted in a digital hinterland, nor a tool left to rust through lack of use.
When I reflect on why I’ve seen integrated AI work so much better—both at Omnitas and with clients—three factors stand out:
- Everyone is lifted: Not just the techies or the risk-takers, but the backbone of your business—the client managers, project leads, support pros.
- Outcomes are measurable and improved: Hours are saved, quality goes up, and feedback loops (from both people and process) spur continual improvement.
- Culture shifts (for the better): AI is no longer scary, nor is it a burden. It simply lets people do what they do best—while the machine lifts the grunt work, reliably, in the background.
The Bottom Line: The Right “Why” for AI
Ask yourself:
Are we really getting the most out of this technology? Have we made AI work for us—or are we still working for the software vendors?
The difference isn’t the tool, and it’s not even the people. It’s the system—the DNA—into which that tool is woven.
It’s time for the shiny hammers to be put to use within solid blueprints. That’s how smart businesses set themselves apart. And for those ready to move beyond experimenting, I’ll say this: the door to true efficiency, compliance, and “unfair” competitive advantage is never as far away as it seems—if you know how to build AI into your business, not just your toolkit.
