From the AI for SMEs panel at the Second German-Irish Business Forum 2026. The three patterns Irish SMEs keep hitting on AI adoption — and why the strategy work usually comes second, not first.
Yesterday I was on the AI for SMEs: From Confusion to Competitive Advantage panel at the Second German-Irish Business Forum 2026, hosted by the German-Irish Chamber of Industry & Commerce. Mark Kelly (AI Ireland), Aarthi Kumar (CK Delta) and I spoke under the moderation of Michael Dowling at DCU Business School. The hour produced more agreement than the format typically allows, particularly on a question that doesn't usually get asked: what are Irish SMEs actually stuck on?
The honest answer, across the room, was that most are not stuck on AI strategy. They are stuck before it. Three patterns kept surfacing, and they are worth setting out plainly because the conversation about Irish SME AI adoption has so far been dominated by the wrong starting point.
Pattern 1: Starting with the tool, not the diagnosis
The first instinct for most Irish SME owners is to choose a tool. Copilot or ChatGPT, Gemini or Claude, an AI agent platform or a workflow automation. The selection happens, training gets organised, the rollout completes — and twelve months later the productivity has not moved.
The pattern is consistent because the starting point is consistent. SMEs treat the question "how do we adopt AI?" as a technology question. It almost never is. The work that creates the value isn't the tool selection. It's the diagnostic that identifies where senior time is being consumed by tasks that should not require a senior person, where existing systems are being underused, where manual steps in cross-team handovers are absorbing hours that no one is tracking.
In a recent Acuity engagement with an Irish debt recovery law firm, the diagnostic surfaced 1,525 hours of senior solicitor time being lost annually across seven categories of manual work — none of which required a lawyer to perform. The firm had been about to scale headcount to handle a new Revenue Commissioners contract. After the audit, they didn't need to. None of that work emerged from a tool. It emerged from process auditing that should have happened before any tool conversation began.
The first question for an Irish SME considering AI is not "which tool?" It is "where are our hours actually going, and what would we recover if those hours came back?"
Pattern 2: Over-procuring before configuring what's already there
The second pattern is closely related. Irish SMEs routinely buy new AI tools without first auditing what is already built into the systems they pay for. The most common example is Microsoft 365 Copilot — frequently introduced as the AI initiative when the existing M365 environment is itself substantially underused. SharePoint configurations that would automate document workflows are not set up. Power Automate is licensed but not deployed. Default templates and automations sit untouched. The productivity opportunity inside the existing licence is often larger than the productivity opportunity from the new add-on.
The discipline this pattern requires is uncomfortable. It involves an audit before a purchase, configuration work before a deployment, and an honest conversation about whether the existing tools have been configured to do what was promised when they were procured.
The same logic applies to CRM systems, financial software, document management platforms, and HR tools. The AI features inside these systems — embedded summarisation, embedded classification, embedded automation — are usually under-discovered. Configuring them produces gains that no new tool purchase will replicate, because the data is already there and the workflows are already there.
The principle: configure what you own before procuring what you don't.
Pattern 3: Treating AI as a future project rather than a current cost
The third pattern is the most strategically expensive. Irish SMEs talk about AI as a project they will get to. They will assess it next year, plan it in Q3, scope it after the budget cycle. Meanwhile, the cost of not having an AI-informed view of operations accumulates quietly every week.
An Irish food retailer Acuity worked with was operating on margin without understanding why. The business had not procured AI. The diagnostic — informed by AI-assisted analysis of supplier terms, product mix, and pricing — identified €560,000 in recoverable margin within seven days. Supplier renegotiations followed. SKU rationalisation followed. Pricing corrections on items priced below cost followed. No technology was deployed. The work was finding what was already there.
The cost of deferring this kind of work is not "what we paid for AI next year minus what we paid this year". It is the margin that wasn't recovered, the hours that weren't returned, the operational visibility that wasn't surfaced. Those costs do not appear in any budget line. They are absorbed quietly into the operating performance and disappear.
The framing question for Irish SME owners isn't "when will we adopt AI?" It is "what is the cost — this quarter — of not having an AI-informed view of how our business actually runs?"
What the panel agreed on, and what the AI Ireland and CK Delta perspectives added
Mark Kelly from AI Ireland framed the SME conversation around the gap between AI literacy and AI capability — a gap that the Article 4 obligations under the EU AI Act formalise but do not solve. Aarthi Kumar from CK Delta emphasised the operational practicality of AI for SMEs being held back by the absence of clean data to work with — not by the absence of tools. Michael Dowling at DCU Business School made the observation that the academic and the practitioner views of AI adoption are now converging much faster than they did with earlier technology cycles.
The convergence point across the panel was the same as the pattern above. The strategy work is not what stops Irish SMEs from adopting AI. The work that precedes it is. The diagnostic, the readiness, the configuration audit, the cost-of-inaction analysis. These are not glamorous. They are the work.
What this means for Irish SMEs in 2026
A practical starting position for any Irish SME considering AI adoption, after the German-Irish Business Forum conversation, looks like this:
Before procurement, do an honest diagnostic. Where is senior time actually going? What is consumed by tasks that should not require a senior person? What manual steps in cross-team handovers are absorbing unmeasured hours? The output of the diagnostic — not the output of a tool catalogue — should drive what comes next.
Audit the configuration of existing tools before buying new ones. Microsoft 365, the CRM, the document management system, the finance platform. Each has embedded automation and AI capability that is almost always underused. Configuration work produces gains that new tool procurement does not.
Treat the absence of AI-informed operational visibility as a current cost, not a future project. The margin you are not recovering, the hours you are not measuring, the supplier terms you are not interrogating — these are happening this quarter. They will happen next quarter too if the work doesn't.
The forum yesterday made one thing clear: the gap between Irish SMEs that adopt AI well and those that do not will be set by what they do before the AI strategy conversation begins. Not by what they do after.
The Second German-Irish Business Forum 2026 was hosted by the German-Irish Chamber of Industry & Commerce at the Herbert Park Hotel, Dublin, on 26 May 2026. The full agenda and participant list is available here.