Why AI Success Depends on Redesigning Workflows, Not Buying More Tools

Companies are under pressure to adopt generative AI, and many are rolling out tools to staff at speed. According to S&P Global Market Intelligence, 42% of companies abandoned most of their AI initiatives in 2025, up from 17% the year before. The average organisation scrapped nearly half its AI proofs of concept before they reached production. Yet many boards are still pushing for faster adoption and quicker returns. Technology may not be the problem.
Morten Halvorsen, chief revenue officer of Procano, an IT management firm serving businesses across the Nordics, sees a common mistake. He believes that many leaders treat generative AI as an isolated gadget rather than a reason to redesign how work gets done. In his view, buying tools without rethinking processes wastes capital and creates fragmented systems. "Companies are investing in the tools, not in the people or the processes," Halvorsen says. "That's where most AI projects go wrong."
Many companies may buy licences and hand out access, expecting productivity to rise. Halvorsen says staff often receive little training, and companies rarely build shared knowledge bases the AI can draw on, such as brand materials and templates. In his experience, many employees end up using AI for one-off tasks rather than rethinking how their work is done. He adds that tools without context or training rarely change how a business performs.
Halvorsen notes that enthusiastic employees often lead AI adoption on their own. They tend to automate their own pain points, or those of whoever complains loudest, rather than company-wide priorities. As a commercial leader, he would naturally start with sales and marketing, when IT production would deliver about ten times the impact. "If I were to become the AI superhero, I would always start where I'm the most hurting," he says.
Another mistake is building automation around outdated systems. In Halvorsen's view, companies should start with the outcome they want and work backwards, even if that means replacing old tools. He also warns that time saved through automation can quietly disappear if there's no plan for how it will be used. "Efficiency gains need a destination," he says.
Procano's approach, Halvorsen explains, starts by mapping processes department by department to find where the most time is lost. Rather than giving AI to everyone at once, he recommends starting with the departments where the largest number of people spend time on repetitive tasks. Objective business analysts work alongside internal champions to identify company-wide gains, not individual frustrations. He adds that each automation needs a defined outcome, whether that's reallocating people to higher-value work, increasing output, or changing roles.
For Halvorsen, soft policies alone don't stop staff from using unapproved AI tools. If employees find approved tools inadequate, they may turn to their own, creating "shadow AI." He recommends governance software that enforces policy, alongside giving staff tools they actually want to use. "Control and usability need to work together," he notes.
Halvorsen expects routine IT support to change drastically by 2030. "Automated platforms will handle most routine service desk requests, such as password resets, with instant responses at any hour," he says. He believes customers will increasingly buy a support platform rather than support staff. He points to out-of-hours support as an example, where businesses often pay a premium for on-call staff or rely on offshore services to cover the gap. In his view, providers that don't adapt may become uncompetitive.
That shift has implications for buyers too. Halvorsen cautions leaders against signing long multi-year IT service contracts today. "Committing to an IT contract longer than three years right now is a serious risk," he says. "The services you're paying for may look completely different by 2030." He elaborates that a contract running into the mid-2030s could mean paying manual-labour prices for work that may be automated.
Halvorsen's message is that AI strategy is not a software purchase, but a decision about processes and people. Leaders who map their workflows and plan how roles will shift may see real returns. For mid-sized businesses in particular, he believes those that act now are likely to be better placed than those waiting for the technology to settle. In his view, if a roadmap has no plan for how human roles change once time is saved, the automation may already be failing.
© Copyright IBTimes 2026. All rights reserved.

