Photo by Giulia Boratto
On Thursday, I attended the PMI Innovation Forum at CSCS, Switzerland's National Supercomputing Centre, on AI for SMEs, with insights from Dr. Maria-Grazia Giuffreda and Dr. Victor Holanda Rusu of CSCS, and Andrea Barni of Fondazione AGIRE | Tecnopolo Ticino.
Representatives from Canton Ticino and IDSIA USI-SUPSI were also present, adding further perspective to the discussion.I left with a few observations worth sitting with. A starting point from the #OECD framed the room: 61% of surveyed SMEs say they have adopted AI. Of those, 76% are using off-the-shelf tools with no strategic framework. And only 21% report significant or transformative impact.So what is actually happening in that gap?
A few reflections, some raised during the forum, others shaped by my own reading of what I heard and discussed during the final exchange with other participants:
· 𝐓𝐡𝐞 𝐯𝐨𝐜𝐚𝐛𝐮𝐥𝐚𝐫𝐲 𝐢𝐬 𝐨𝐟𝐭𝐞𝐧 𝐦𝐮𝐜𝐡 𝐧𝐚𝐫𝐫𝐨𝐰𝐞𝐫 𝐭𝐡𝐚𝐧 𝐭𝐡𝐞 𝐭𝐞𝐜𝐡𝐧𝐨𝐥𝐨𝐠𝐲. Many conversations still stop at chatbots, while the field already includes LLMs, SLMs, RAG architectures, multimodal systems, and increasingly agentic applications.
· 𝐓𝐡𝐞 𝐨𝐛𝐬𝐭𝐚𝐜𝐥𝐞𝐬 𝐚𝐫𝐞 𝐧𝐨𝐭 𝐣𝐮𝐬𝐭 𝐭𝐞𝐜𝐡𝐧𝐢𝐜𝐚𝐥: maintenance, hardware, and training costs; skills gaps and lack of time; limited cybersecurity readiness; and the sheer speed at which tools and offerings are evolving, which makes it difficult for organizations to keep pace. 𝐓𝐡𝐞𝐲 𝐚𝐫𝐞 𝐚𝐥𝐬𝐨 𝐨𝐫𝐠𝐚𝐧𝐢𝐳𝐚𝐭𝐢𝐨𝐧𝐚𝐥: cultural, generational, and closely linked to leadership involvement.
· 𝐓𝐡𝐞𝐫𝐞 𝐢𝐬 𝐚 𝐜𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐠𝐚𝐩. Those who have a deeper understanding of what AI can do and those who decide how and for what to use it often do not speak the same language, and that bridge is not being built fast enough.
· 𝐀𝐜𝐜𝐨𝐮𝐧𝐭𝐚𝐛𝐢𝐥𝐢𝐭𝐲 𝐢𝐬 𝐨𝐟𝐭𝐞𝐧 𝐮𝐧𝐜𝐥𝐞𝐚𝐫. In many organizations, it is still not well defined who is responsible for verifying and validating what AI produces. At the same time, reliance on AI outputs is already growing, even though hallucinations, inaccuracies, and biases remain common enough to require careful human oversight. Without clear accountability, output becomes input with no one checking the chain.
As AGIRE put it sharply, 𝑡ℎ𝑒 𝑚𝑎𝑖𝑛 𝑟𝑖𝑠𝑘 𝑖𝑠 𝑛𝑜𝑡 𝑓𝑎𝑖𝑙𝑖𝑛𝑔 𝑡𝑜 𝑎𝑑𝑜𝑝𝑡 𝐴𝐼, 𝑏𝑢𝑡 𝑎𝑑𝑜𝑝𝑡𝑖𝑛𝑔 𝑖𝑡 𝑖𝑛 𝑎𝑛 𝑢𝑛𝑠𝑡𝑟𝑢𝑐𝑡𝑢𝑟𝑒𝑑 𝑤𝑎𝑦. Too often, adoption remains shallow, limited to off-the-shelf tools used for isolated tasks rather than embedded in a broader strategic framework.
𝐏𝐞𝐨𝐩𝐥𝐞 𝐭𝐫𝐚𝐧𝐬𝐟𝐨𝐫𝐦 𝐨𝐫𝐠𝐚𝐧𝐢𝐳𝐚𝐭𝐢𝐨𝐧𝐬, 𝐧𝐨𝐭 𝐭𝐨𝐨𝐥𝐬.
Maybe the real measure is not how many tools we use, but how prepared we are to govern them, explain them, and validate what they return.
#AIforSMEs #DigitalTransformation #B2BCommunication #Switzerland #StrategicCommunication #PMI
