a daily news desk
Tools

OECD: only 3.6% of SMBs have crossed into agentic AI, and customer discovery is one of the gaps

A survey of 2,000+ small firms across 12 countries finds most are still running off-the-shelf tools in isolation while a tiny minority redesign operations around agents.

By Iris Holloway·Tools & frameworks·

The OECD’s 2026 D4SME Survey, synthesized Monday by PYMNTS, finds that 61% of small and mid-sized firms now report using AI, but only 3.6% have crossed into what the authors call the “AI Champion” tier of agentic systems deployed across the business. Published April 13, the survey covers more than 2,000 SMEs across 12 OECD countries, and the headline number isn’t the adoption rate. It’s the gap between adoption and depth.

Of the adopters, 56.6% use AI only for isolated tasks and 19% apply it across multiple functions. A full 76% remain “AI novices” running off-the-shelf tools in isolation. That leaves 96.4% of adopters below the agentic tier where, PYMNTS notes, agents are being pushed into production across four core functions: customer service, supply chain, pricing, and customer discovery.

The OECD describes agentic AI as “primarily used to automate multi-step workflows such as managing bookings, invoices or marketing campaigns,” and profiles a Tokyo micro wholesale firm that built custom agents for multi-language Q&As and project negotiations, reporting higher revenue and time saved on invoicing. That’s the top of the pyramid. The World Economic Forum sized the agents market at $5.4 billion in 2024, projected to reach $236 billion by 2034, with $3 trillion in productivity gains across the decade. Nearly all of that value accrues to firms that get past the novice stage.

Why most don’t: cost, training time, and skills gaps. HBR’s Joseph Fuller argued in March that the shift is about managing work, not installing software, scoping an agent like a new hire with a defined role and escalation path. Founder-led businesses rarely have the bandwidth for that onboarding, which is why agents are becoming the first step in buyer journeys while most owners are still choosing between point tools and the pay-per-lead pricing shift.

A done-for-you service like LemonLime is positioned around that gap, studying the business and preparing finished outreach and content without an internal AI team to run it. The structural read: the 3.6% isn’t a ceiling. It’s a window.

Sources