HiQ’s new AI chief calls for business-linked AI budgets

Anders Kullenberg, HiQ’s new Head of AI, says companies should link AI spending to skills, business value and measurable results instead of applying the same usage cap to everyone.

AI costs have become a practical issue as companies move from experiments to broader use of generative AI tools. According to figures from UBS, about 60 percent of companies have begun tightening AI spending. Swedish companies are reportedly more than 20 percent below the global average in AI returns.

Kullenberg, who has worked with software development, embedded systems, computer vision and autonomous systems at companies including Scania and Volvo, says usage alone is a poor measure of productivity. For example, token consumption should be connected to existing business metrics, such as production bugs, sales per salesperson, or time saved in HR and finance.

He supports differentiated access to more advanced and costly tools, based on training or certification, while ensuring that employees can progress. He also describes a HiQ pilot in which routing requests to models suited to each task reduced costs to one seventh while developers reported that functionality was maintained.

Kullenberg says the same routing layer could support information classification and rules about which data may be sent to which model or environment. He warns that excessive restrictions can drive shadow IT, while buying GitHub Copilot licenses for every developer is not, by itself, an AI strategy. He argues that companies should focus more on integrating their data, which he calls their main differentiator.