Gunda Karnīte

Associate, GenAI, IT Advisory, PwC Latvia

Kristians Leonovics

Associate, IT Advisory, PwC Latvia

Manual procurement documentation is time-consuming and prone to human error. Employees often use previous procurement documentation to create new documentation, which, if not thoroughly reviewed, can harbour risks. According to Gartner¹, 65% of procurement leaders are actively investing in artificial intelligence (AI) to improve productivity rates and reduce the risk of human error. AI is a useful tool for tasks such as analysing historical procurement data, preparing documentation and evaluating candidates. 

The first step is to explore and try out the available AI tools. The tools are varied and available in three price categories: free, subscription and pay-as-you-use. We recommend that you consider a subscription when shopping. Free versions are not suitable as they do not have essential features such as agent creation, past data storage and unlimited prompts that are available in paid tools. Subscription MI tools such as ChatGPT, CoPilot, Claude, etc., are easily available for USD 20 - 30 per user per month.

The second step is to configure the AI tool to perform the required functions in the procurement process. This includes creating trained agents that receive information such as sample purchasing forms, historical company data and operating instructions, as well as a detailed description of the desired outcome, constraining factors, data formatting and indirect context such as geographic data, seasonal changes, supply chain specifics or internal company conditions that are not available to the AI agent.

The third step is the evaluation of sourcing candidates. You can also set up a specialised agent for this task by providing evaluation criteria, information forms submitted by the applicants and detailed work instructions as guidance. Such an agent must be able to compare and evaluate applicants not only based on the documents submitted, but also to obtain additional information from public sources on the Internet. It must carry out a comprehensive assessment of applicants, taking into account other projects undertaken, competencies and reputation.

It is important to introduce the AI tool gradually, with research and testing, followed by configuration and training, through to implementation and ongoing optimisation. It is important that expectations are realistic and that AI is an efficiency improvement tool that speeds up the completion of large parts of the work. However, the result must always be reviewed and improved.