Most firms say they're good at AI. Few have changed how the work gets done. Map the work first

The first is a reality check. Avanade's Build Beyond Limits: AI Reality Report, reported on 3 November, found that 80% of UK organisations describe themselves as established, advanced or AI-first. Yet only 17% have fundamentally redesigned their operating model around AI, 61% admit AI hasn't yet delivered financial returns that justify the spend, and 65% have never formally calculated AI ROI with finance or leadership. Most candidly of all, 71% admit to portraying themselves as more AI-mature than they really are.

The second is a sensible answer. Morrisons has partnered with Duvo.ai, and the project doesn't start with a chatbot or a big licence. It starts by capturing how processes actually work across the grocer's people, teams and systems, then deciding which repetitive tasks to simplify, redesign and automate.

That order of operations is the lesson for every UK business, whatever its size.

The maturity gap, in plain terms

Avanade surveyed large organisations (350 in the UK, all over $500 million in revenue), so these aren't firms short of budget or talent. Their problem isn't access to AI. It's that AI has been added on top of existing work rather than changing it:

• 69% say AI investment is still focused on incremental improvements rather than transformation.
• 79% haven't fundamentally reshaped how teams operate.
• 42% have a documented AI strategy that isn't consistently executed.
• 97% have hit challenges deploying and scaling AI.

The most commonly cited transformational benefits were operational efficiency and customer experience improvements, named by 23% of respondents. Revenue and cost reduction were the least affected. Unsurprisingly, 45% now say finding the use cases with the best return is among their highest priorities.

If big firms with dedicated AI teams are finding this hard, smaller businesses shouldn't feel behind. They should feel warned, and they have one big advantage: their processes are small enough to map in days, not months.

What Morrisons is doing right

Morrisons' Group Productivity Director, Gordon MacPherson, put it well: the grocer's processes "have evolved over decades", and understanding them in detail means automation decisions "can be based on evidence rather than assumptions".

Duvo's co-founder, Tomáš Čupr, framed success in refreshingly concrete terms: "an invoice matched, a dispute resolved with the evidence attached, or the correct result written back to a core system." He also argued that AI should work inside the systems teams already use, with human oversight, "rather than becoming another layer of disconnected pilots".

Three ideas worth borrowing:

1. Evidence before tools. Find out how the work really flows, including the workarounds, spreadsheets and email chains nobody wrote down.
2. Outcomes you can count. Invoices matched, tickets resolved, orders processed without rekeying. Not "AI usage".
3. Write back to the system of record. AI that drafts something a person then copies into another system has only moved the work around.

A one-week version for UK SMEs

You don't need an enterprise programme to apply this. A focused week is enough to get started:

• Day 1: pick one process that hurts. Order to invoice, enquiry to quote, supplier bill to payment, or returns to refund. Choose one with volume and repetition.
• Days 2–3: map it as it really runs. Sit with the people who do it. Note every system touched, every copy-and-paste, every approval and every place work waits.
• Day 4: mark the candidates. For each step, decide: remove it, simplify it, automate it with rules, or hand it to AI with a person approving. Many of the best wins need no AI at all.
• Day 5: set the baseline and the business case. Time per item, error rate, items per week. That's the number finance will ask about, and the one 65% of large firms never calculated.

Then build the smallest working version, connected to the real systems, and measure it after six to eight weeks.

The optimistic bit

Support is growing. The Scottish Government convened industry leaders with techUK in Edinburgh today, pointing to estimates that AI could add more than £23 billion a year to Scotland's GDP by 2035, and its £1.8 million National AI Adoption Programme for SMEs is already helping smaller firms get started. Across the UK, the appetite is there.

The firms that turn that appetite into returns won't be the ones with the most AI tools. They'll be the ones who understood their own work well enough to know exactly where AI belongs.

Why this matters for Clarity Growth

Clarity Growth helps UK SMEs and scale-ups build connected systems, where websites, ecommerce, CRM, finance and operations share clean data, and automation and AI are added where they measurably help. Process mapping is usually our first step: understanding how work actually flows today, choosing the system of record, then connecting the tools a business already owns so AI can act on real data and write results back, with sensible approvals and logging. It's the Morrisons approach, scaled to a business that wants measurable results in weeks rather than a multi-year programme.

Sources

1. UK firms overstate AI maturity as transformation lags, Avanade research finds — Digital Infrastructure (3 November 2026; Avanade Build Beyond Limits: AI Reality Report, Vanson Bourne, Aug–Sep 2026, 350 UK respondents at organisations over $500m revenue: 80% self-rated advanced, 17% redesigned operating model, 61% no justifying return, 65% no formal ROI review, 71% overstate maturity)
2. Morrisons partners with Duvo as UK grocery giant eyes opportunities for AI adoption — Retail Technology Innovation Hub (3 November 2026; process capture first, quotes from Gordon MacPherson and Tomáš Čupr)
3. Businesses urged to seize AI opportunity — Scottish Government (3 November 2026; techUK roundtable, £23bn GDP estimate by 2035, £1.8m National AI Adoption Programme for SMEs)