Redesign the work: why the mid-market's AI gap is an operating-model problem

The AI gap is now a size gap

Ireland's Central Statistics Office asks enterprises with ten or more people whether they use any AI technology. In 2025, 20.2% said yes, up from 8.1% in 2023. The headline hides a split by size. Almost six in ten large enterprises (57.7%) used AI, compared with 28.6% of medium-sized and 17.2% of small enterprises¹.

The European picture is almost identical. Eurostat reports that in 2025, 17% of small, 30.4% of medium and 55.0% of large EU enterprises used AI². Among enterprises that had considered AI and decided against it, the most common reason was a lack of relevant expertise (70.9%), ahead of unclear legal consequences (52.5%) and data-protection concerns (48.8%)².

The gap is widest exactly where the money is. Using AI to draft an email is common. Using it to run a workflow, such as quoting, scheduling or claims, is where the size split is sharpest: 4.3% of small Irish enterprises against 28.6% of large ones¹.

Large firms are pulling further ahead

McKinsey's 2026 global survey of 1,719 executives points the same way. Fifty-four percent of respondents from organisations with at least $1 billion in revenue said AI was scaling across their enterprise, against about a third of those from smaller organisations. The share of large organisations scaling AI agents rose from 27% to 40% in a year. Among smaller organisations it stayed flat at 22%³.

Taken together, the official statistics and the survey describe a market where the largest firms are building a second operating advantage on top of the ones they already had: data, systems and specialist teams.

But scale is not the same as value

The same survey has a more useful finding. Only 37% of respondents attribute any EBIT impact to AI, essentially unchanged from 2025. The group McKinsey calls high performers, who attribute at least 5% of EBIT to AI and describe its impact as significant, is still about 6% of respondents³.

What separates them is not the tools. Nearly three-quarters of high performers say they have fundamentally redesigned workflows because of AI, against about a quarter of everyone else. McKinsey describes them as firms that "fundamentally redesign workflows that are enabled by AI rather than insert AI into existing ones"³. In its 2025 survey, high performers were also more likely to have defined processes for when a model's output needs human validation, one of the factors that most distinguished them⁴.

Why the middle of the market is better placed than it looks

This section is our view, not survey data.

• One workflow is a large share of the business. In a regional installer, distributor or service firm, the path from enquiry to quote to booking often carries most of the revenue. Fixing it is not a pilot in one department. It changes the firm.
• The decision maker is in the room. High performers stand out for senior leaders who own AI initiatives³. In an owner-run firm, the owner is the senior leader.
• Fewer systems to unpick. A mid-market firm usually runs on a phone, an inbox, WhatsApp, a spreadsheet and an accounting package. That is messy, but it is far easier to redesign than twenty years of enterprise software.
• Building is getting cheaper. Thirty-two percent of McKinsey's respondents said their organisation had decided against buying at least one software product or feature because it could now be built in-house with agentic coding tools³. Software that once needed an IT department can now be specified, built and tested in weeks.

What "AI-native" means in practice

An AI-added firm gives its office staff a chatbot and keeps the process. An AI-native firm redesigns the process around what software now does well, and puts people at the points where judgement, safety and trust matter. In a home-services firm, for example:

Five rules for choosing the first workflow

1. Pick a workflow customers feel. Delay that customers complain about in public is evidence that the workflow matters and that a fix will be noticed.
2. Pick one with a number attached. Hours to quote, share of enquiries answered, conversion within 30 days. If you cannot measure the before, you cannot prove the after.
3. Define the human checkpoints first. Decide where a person must approve before anything goes to a customer. This is what lets a firm move quickly without risking its reputation.
4. Design for data protection from the start. Know what personal data the workflow touches, where it is stored and who can see it. Under the GDPR, the firm remains responsible for it.
5. Build small, test with real customers, then scale. A working system used by twenty real customers teaches more than a strategy document read by twenty managers.

On buying help

MIT's NANDA initiative reported in 2025 that buying from specialised vendors and building through partnerships succeeded far more often than internal builds. The study drew on 150 interviews with leaders, a survey of 350 employees and an analysis of 300 public AI deployments⁵. Success there was largely as reported by the companies themselves, so treat the exact ratio with care. We are an advisory firm, so we have an interest in that finding and you should weigh it accordingly. Our own view is narrower. Whoever does the work, the redesign has to be owned by the people who run the firm, or it will not survive the first busy month.

Questions for the next leadership meeting

• Which single workflow carries most of our revenue, and how long does it take from first contact to a price?
• Where do customers complain about us in public, and what step does the complaint describe?
• If we redesigned that workflow around software, where would a person still need to decide?
• What number would tell us, in 30 days, that it worked?

Cite as: Tall Poppy Advisory, “Redesign the work: why the mid-market's AI gap is an operating-model problem”, 3 November 2026, https://tallpoppyadvisory.com/insights/redesign-the-work/. Corrections: hello@tallpoppyadvisory.com.