What public procurement officials want to know when buying GenAI

GenAI is everywhere. Not only in discussions at every procurement event, but increasingly as part of any software or service we are using. It feels like new guidance pops up daily around how to leverage GenAI for the public sector, with dozens of resources alone dedicated to supporting public officials with buying GenAI technology. 

Yet even with all this material out there, we still hear from our partners that they need help. Procurement officials want to better understand GenAI technology and how procuring GenAI is different, and what they should pay attention to. That’s why we’ve set out to develop actionable, plain language, and vendor-neutral guidance for procurement directors and officials for buying GenAI. 

To help us better understand the gap and co-create pieces of the guidance, we recently held two workshops that brought together individuals from the national and state levels, the private sector, nonprofits and academia. These workshops were held in collaboration with the George Washington University Law School Government Procurement Law Program in Washington, DC, and at the Organisation for Economic Co-operation and Development (OECD) in Paris. 

Here are three of our key insights: 

• Good GenAI procurement is good IT procurement. At its core, successful GenAI procurement isn’t a radical departure from existing practices, but an extension of good IT procurement. For example, good IT procurement practices include focusing on desired outcomes rather than detailed technical requirements, comprehensive market research, strong vendor management, and flexible delivery approaches. These practices are all more important than ever to get right for GenAI procurement. While GenAI does bring some unique considerations around risks, if the government is already set up for outcome-driven agile procurement, then most of these aspects can be incorporated into existing procurement frameworks.

• Procurement teams, buyers, and technologists need to work collaboratively together across agencies to get the best outcomes. Procurement can have influence and some decision-making power in the process of buying or building GenAI solutions. But procurement can’t do their job well without good policies and people in place to support a strong decision-making and implementation process. Procurement and non-procurement stakeholders should understand when procurement is responsible for decisions, and when it is their job to ask smart questions. Procurement teams need to be ready to ask and check that good practices for both tech deployment and procurement are followed.

• Practitioners have a strong appetite to fill the GenAI knowledge gap with case studies and hands-on training. We heard that people really want case studies. But while there are a lot of good examples of using GenAI in the public sector, a lot less has been documented around how this technology was purchased and how existing procurement methods and contract vehicles can be applied. In addition, while lots of guidance documents, frameworks, and policy papers exist for GenAI, there’s a critical lack of hands-on training to help procurement teams put these insights into practice.
  
  A few governments stand out for their leadership in thinking strategically about GenAI and procurement – such as the national government of Australia, with their national framework for the assurance of artificial intelligence in government, or City of San Jose, California, which leads the GovAI Coalition. But overall, we heard that governments have a lot of room to build skills and strategies.

We will now take these insights and more from the workshops forward as we develop the next iteration of the guidance, with an estimated public launch in October. 

Do you want to share a case study around buying AI technology? Please reach out to me at Kaye Sklar.