Your browser doesn’t support HTML5 audio
In common with enterprise technology departments, procurement teams will be impacted by AI and play a role in its adoption. Is the procurement function prepared for the arrival of AI?
Procurement shares much in common with the finance and legal arms of business. All three are process, document, and information-heavy. Finance and legal have been regularly cited as ripe for AI transformation — and what about procurement? Melanie Humphries, CEO of Horizon Seven, a sourcing and outsourcing specialist consultancy that works with organizations looking to make procurement decisions or improve procurement processes, says:
AI could help at the beginning and the end of a procurement process. At the beginning, it would be useful for the research and heavy lifting to work out who the suppliers are that you could work with. At the other end, contract writing could use AI to speed up that element.
CTO Keith McFarlane of AI procurement technology provider Globality agrees:
Much of the improvement is in the scoping of suppliers; AI is a tremendous fit for scoping.
Globality claims that between 70 and 80% of the questions organizations need to ask in scoping for suppliers can be determined by AI, which reduces waste in a request for proposal (RFP) document. He adds:
Generative AI gives us the ability to reason what are the best questions and determine the 20 to 25 questions to ask so we have a more efficient process.
Other areas of procurement that AI could improve include linking and extracting data from databases, trend analysis, automated scoring, creating visualizations, and identifying special value or ESG target opportunities.
None of this removes the need for highly skilled procurement staff, Humphries says:
The tools can get rid of a lot of heavy lifting, but that last 20-25% requires the human to deal with all the nuances that only a human would understand. So you would get the benefit of time efficiencies on the more administrative tasks.
Recent advances in agentic technology are not only allowing pure-play procurement vendors like Globality to enter the market, as we reported late last year process and payments leaders such as Coupa also have procurement in their sights.
AI risks
As with other areas of the enterprise, there are risks to using AI in procurement. Relationships between service providers and the enterprise are at the heart of the procurement process, and the ability of AI to hallucinate and misrepresent facts is concerning. That risk is exacerbated by procurement, like many other areas of the organization, not having its data in a state that would enable a trustworthy use of Generative AI. Humphries says:
There are a lot of companies out there trying to sell an AI tool that they say is all an organization needs to do a contract sweep, but as with anything, the information has to be there, and most organizations don't know where all their contracts are. So you have to be ready for AI.
McFarlane says the question of data quality comes up frequently when the procurement technology company talks to CFOs and prospective customers. Organizations are also worried that their contract data is placed into a large language model (LLM) and is no longer part of the intellectual property of that business. The Globality CTO says:
We do not fine-tune LLMs using customer data or allow LLMs to train themselves on customer data. For our internal modelling that our data scientists use, they are using customer data by agreement for classification models.
Globality counts global financial service firm Fidelity, the UK's retailer Tesco, and telco BT as customers.
The other risk is that procurement teams, at present, lack the skills to both procure AI and use it effectively. In her observations, Humphries of Horizon Seven says:
Change means time to learn and adapt, and they don't have time for it, so AI sits in another one of those buckets of 'I know I should embrace it and understand it, but I have 25 projects and all these objectives to save money, so I just don't have the capacity.
This is not restricted to AI, she says. Horizon Seven often works alongside procurement teams that have been tasked with major deals, such as the selection of a cloud provider, and the category manager has no prior experience.
At last year's AI Commercial Lifecycle and Procurement Summit, one procurement leader pointed out that without skills, procurement teams will not gain much from adopting AI:
If you are copying and pasting standard requirements into AI, you will only get standard responses.
Transform first
Adopting AI will just add another technology to procurement. Digital leaders in technology, finance, and procurement have all voiced to me that the role needs to undergo significant modernization. Humphries sketches out the situation:
Traditional procurement functions are typically over-worked, under-resourced and under-informed. That is not because they are not professional; it is because of the nature of where they sit in the organization.
When procurement staff sits within business lines, she finds that there is better alignment and, therefore, results. Sadly, in most organizations, procurement is one of the responsibilities of the CFO and can often become only focused on cost savings. The alignment issue has been one that dogged technology teams for a number of years. CIOs have since moved to a cross-functional team way of working, which has gone a long way to address alignment issues.
Even if procurement teams undergo significant transformation and adopt AI, the role is an area of organizations that requires significant human input. These include setting up and attending workshops with potential suppliers and decision-making. In addition, procurement teams have to be the guardians of good governance in organizations. When procurement governance breaks down, it can be toxic for an organization. One example that will be familiar to UK readers was the subversion of rigorous public sector procurement practices during the COVID-19 pandemic by ministers from the governing Conservative Party.
Are you ready
It is not only procurement teams that need to be ready for AI. Humphries says technology vendors and other suppliers will need to prepare their data for AI access. She says:
When AI gets embedded into a buying process, they are going to be screened by AI, so are they presenting themselves and their information in a positive way? We did some research using chatbots to see what they could harvest from suppliers, and the information was not good enough to drive a decision. I don't think supplier organizations have worked out that this is what procurement people are going to do.