AI in procurement: Lessons learnt with Ninian Wilson, CEO of Vodafone Procurement Company

José Oliveira: I think what is very important is how AI is boosting existing tools and existing solutions, as opposed to what it is actually generating by itself.
In the procurement space, it’s making a huge change in the way it allows us to process data, to enrich it, to make it more relevant, to make it better quality, and to connect it. It's much faster to read contracts these days, it's much faster to understand what's in invoices, it's much faster to combine them, and it's also much faster to get high-quality synthetic market research data. So, AI is doing a lot of that by itself.

What it can do too, although I think it is lagging a little bit behind, is empower existing solutions. The most visible thing in procurement is the S2P suites – the ERFX e-sourcing, the P2P processes, and so on. Where AI can most help is by redesigning those processes and making them much less flow-based and much more conversational for the user. But that requires a lot of work in terms of implementation and adapting those rules to your own needs and processes before it becomes a less risky proposition.

In terms of “Make” versus “Buy”, if you ask large organizations how they approach it, waiting to buy for them is not an option, right? They understand that there are several use cases already there, they already have an instinct that it's doable, and that they should press the button on it and move with it. I think there are very good reasons to think that it is possible to stick with “Make”.

I think it is also important to look at the fact that there is going to be a lot of disruption in the vendor market, and therefore the “Buy” option might be very exciting. But it is still a bit of a question mark in terms of who is going to be around in 12 or 18 months. And do the implementations work as well as they look, and how well do they integrate with other aspects? So there’s a bit of a hold-and-see attitude – maybe wait a little bit for some leaders to emerge. We don’t yet see a category killer on the Buy side, and that’s the challenge. Eventually, Buy will make sense, but there’s a lot you can already do through Make.

Ninian Wilson: Developing yourself, you're going to have to have some core capability to do that. But there are companies out there who'll offer to do things for free of charge for you, so you can learn without having to buy immediately, and I'd encourage everybody to do a little bit of learning. Speak to some of the companies that are out there and get them in to come and show you what they can do, so you can learn a little bit, and then you can decide whether you do the Make or Buy decision, once you’ve got a better feel for the technology and how it can transform your function.

José Oliveira: I think there's some immediate areas of opportunity that we're seeing. Contracting is an example, because there's a lot you can basically read and make sense of and triage and prioritize and analyze much faster. 

There’s an expectation that processes will become much more fluid and conditions-based, but I don’t think we’re there yet. Things around baselining cost forensics information are perceived as also high-potential, but the problem there is more around the habits that people have in consuming that information. I can give you a lot of information now that I wasn't able to do even six months ago, but you're still not going to consume it the right way. The challenge is still the adoption and the creation of habits around how you use these things.

And the second challenge, more for a procurement leader, is: yes, we can focus on what we do right now, and that's going to be accelerated – but also there should be a lot of focus around what we can do that we were not able to do before. If you look at how your team should look like in 12 months, they should probably be doing much more research, much more analysis than they do today, which is currently more related to document production and compilation.

Ninian Wilson: Really, really good points. The way I look at it, I think a procurement process is fairly standard. You want to understand what's out there in the market, you want to gather internal requirements, you want to put some sort of commercial documentation together, and you want to run some level of competition. I think there are various points where AI can really help reduce the admin. There's no reason an AI agent can't write the tender. That’s simple stuff. If you've got a big database of what you've done before, there's no question it can do that. 

We've also got our autonomous sourcing system giving coaching and feedback to suppliers through the tender process. We used to do this manually – telling suppliers, for example, “You’re not very strong on health and safety” or “Your technology roadmap or commercials need improvement” and offering suggestions on how to get better. That dynamic feedback is something AI can now deliver.

The contracting piece: 100% agree with Jose. You know, we do 2,500 to 3,000 contracts a year, we've got a database that's 30,000. We know exactly where we give, we know where we don't give. We can give that to AI to negotiate. But then, you might not use this tooling on your strategic partnerships, right? It might be more of the simple procurement you're doing with AI, to start off with. And obviously, the thing that is the key differentiator, as Jose has said, is that context around your cost base, and how you analyze that and get insights. So, I think it can really help the function. We think we'll see about 30% reduction in cycle time for a tender.

You’ve got to remember, on the flip side of this, our supplier partners are already using AI to write their response documents. So, in a sense, it’s becoming AI-to-AI communication – discussing and proposing. You can look at that two ways: you can be a little nervous, or you can be excited. I’m excited – but José’s point about change management and helping people build new habits is critical. Because I know the thing procurement people like doing the least is research, and the thing they love doing most is sitting down to negotiate. So you’ve got to reshape the DNA a bit – get people thinking differently and create those new habits. I’ve met so many people who just fly into a negotiation without doing the groundwork. AI gives people the ability to do more of that research. The question then becomes: how do you create a commercial advantage if everyone’s using the same tools?

Ninian Wilson: We're looking at a return in excess of 20 to 1 in that capability, in terms of additional value delivered back to the company. Some of that's around speed, but you've also got to remember, all of our supplier partners are implementing AI. So there's efficiency within the function, but there's also significant efficiency and value being created in the supply base, and as procurement people, we'll want to tap into that as well.

I think there is a tendency to say AI is going to solve everything, but some of that for us is RPA – very simplistic robotics – which we're using to improve quality. So, I think it's a little bit like crypto and blockchain. Everybody jumped to it and said it will solve everything, including hunger and all this sort of stuff. You've got to think about where you apply technologies for the best bang for the buck, and you might apply very simple technologies in some areas, and then you can use AI to get more value where it can be deployed best. It’s a bit of a balance.

I think we're in a fortunate position. We've got a single ERP system across our whole company, so we've got a single version of the truth sitting there. The data is probably still nowhere near perfect, but we are in a good position versus a lot of other companies. You've got to have some of the basics, because insights from your own data can be as good or as bad as your own data is.

José Oliveira: We’re seeing, because of the ability to process and to basically tag data points in much more precise ways, it's much easier in a post-contract environment to actually track benefits – to do things like SKU rationalization. So that's quite a lot of work we're doing right now, quite intensely. AI accelerates the precise categorization and actually brings it two or three levels deeper than where it was. It can also use public data to do that tagging – for example, catalogs from suppliers.

Right now, you know, reading invoices is very easy, reading contracts, working around that is particularly powerful, and it's one of the big first quick wins we see with our clients. So, getting much faster to the answer about what am I buying, how am I buying, and also where are opportunities, thanks to importing things into databases, reading them faster, understanding patterns … It's been quite powerful, and high returns.