New blueprint released for AI procurement

A radical overhaul of how cities procure AI and emerging technologies is needed in which governments treat procurement as a dynamic, software-like system rather than a rigid, bureaucratic process, says a new whitepaper.

Released by the Institutional Architecture Lab (TIAL), the report, authored by Leonardo Quattrucci, Adjunct Professor at Sciences Po and Senior Fellow at the Centre for Future Generations, arrives as cities grapple with vendor lock-in, ethical AI risks, and technological change.

Quattrucci, an advisor to governments and tech start-ups, argues that traditional procurement is ill-suited for AI.

“Public procurement is still a brick-and-mortar process, more suited to buying bulks of objects at a given time–like a massive vending machine,” he told Cities Today. “But innovation is different: you don’t know today what is going to be possible tomorrow, even when you are the one inventing the technology.”

Cities often purchase obsolete tech while struggling with shortages of in-house expertise and misaligned incentives. The whitepaper identifies seven systemic flaws, including the “vending machine fallacy” (treating contracts as one-off transactions) and the “hammer-nail problem” (applying the same rigid process to everything from cloud computing to vaccines).

The framework proposes four principles to modernise procurement:

• Adaptability: Shift from risk avoidance to risk management, mirroring agile software development.
• Agility: Use feedback loops, like Australia’s Dynamic Purchasing Systems, to iteratively refine contracts.
• Accuracy: Tailor procurement to the product eg, buying AI like a utility, not a desk chair.
• Accessibility: Simplify processes for start-ups and prioritise user needs, as seen in Norway’s StartOff programme.

Quattrucci emphasises that cities, with their direct citizen relationships, are uniquely positioned to pilot AI responsibly.

“Cities can gain bargaining power in offering responsible experiments to companies in the public interest,” he said.

The research stresses that 62 percent of OECD countries lack professionalised procurement teams, leaving cities vulnerable to vendor hype. Quattrucci calls for hiring “builder” talent–practitioners who can test AI systems–and leveraging data analytics to avoid scandals like the Netherlands’ welfare fraud algorithm.

“Procurement should rebalance information asymmetry,” he noted. “Too often, cities take the blame when AI fails but lack the tools to investigate why.”

The paper urges cities to treat procurement as a strategic portfolio, pool resources with other cities to negotiate better AI contracts, invest in upskilling, and demand algorithmic transparency from vendors.

“AI adoption is not an inevitability, it is a skill,” he added. “Cities can offer tech talent tangible opportunities to build changes and see their effects–which is what motivates builders. What I would ask this cohort of techies to do first is to help cities develop a healthier, more secure, and more integrated data infrastructure.”

Main image: Nicoelnino | Dreamstime.com