Given the supply chain crises and geopolitical challenges within the procurement community, technology has been key in addressing much of the turbulence being experienced almost daily. In particular, AI-powered and predictive freight sourcing software have been a game-changer in helping teams navigate the shift in focus from longer-term and annual tenders to sourcing that is more spot, operational and transactional.
In a recent Procuretech podcast, Keelvar CEO Alan Holland told show host and founder of ProcurementSoftware.site, James Meads, about how Keelvar's eSourcing platform and Intelligent Sourcing Bots can help with the myriad complexities, particularly within the global ocean freight market.
Read a shortened version of the podcast transcript below, or listen to it in full here.
James Meads: Can we briefly introduce Keelvar and who it is you serve?
Alan Holland: Keelvar is a sourcing software vendor, and we emerge from a university research lab. That has informed our DNA, how we approach problems, and how we approach the market.
I used to be a lecturer in artificial intelligence at University College Cork, and I taught intelligent systems to graduate students in the computer science department. Our mission is to help procurement teams scale sourcing excellence: so we're, we're on a mission to help primarily large enterprises apply best practice in negotiations with suppliers, and also scale that so that they can apply it in all negotiations with all suppliers in all spend categories.
JM: I don't think many procurement technology companies came out of a computer science department at a university. How do you see that gives Keelvar a competitive advantage?
AH: The very heart of our competitive advantage is our approach to solving these problems for procurement - it comes from the perspective of a computer scientist.
I'm often asked how I got into this domain. It kind of gets back to side projects I was doing for my parents' company when I was a Ph.D. student. They used to have to interact with procurement teams when they were selling chemicals at a small water treatment chemical company, and I could see from their perspective, that the interactions were quite frustrating.
They had to complete RFQs in the form of spreadsheets, and they never knew which combination of factories or plants they would have to deliver chemicals to, what price or what aggregate volume they'd receive. So they never knew where to place the storage, how much to invest in manufacturing equipment, and so on.
So really, as a computer scientist, I could see this as a data problem and that procurement needed to be collecting much richer information from suppliers. While I was attending some AI conferences, I could see that some computer scientists were looking at the problem of finding more efficient and electronic commerce settings. And it all revolved around allowing bidders to share rich information and package bids, conditional discounts, capacity constraints and such, but that leads to a complicated winner determination problem. And you need algorithms to solve this winner determination problem.
So that's very much at the heart of why we felt that there is a win-win outcome here, because if procurement teams gather richer information from suppliers, suppliers can lower their costs and increase their margins.That was how things got started.
JM: Keelvar has a function that enables optimizing spot sourcing of ocean freight. How can the solution help ensure that the buyer of freight and transportation can get the best out of the market they're facing right now?
AH: Ocean freight is a very good example of a spend category that displays different orthogonal challenges for logistics procurement teams. On the one side, you could have your strategic sourcing events that take a long time to complete. It can take months to negotiate your longer-term contracted rates with ocean carriers if you're a big enterprise.
Our primary competitive advantage over anybody else is that we're the best in the world at supporting very large scale negotiations that require considerable flexibility in the rules around your sourcing process. So you can have great flexibility in terms of the feedback and scenarios you compute at different iterations of the negotiation process.
No matter how big or complicated your negotiation is, we can usually help knock a lot of time off that negotiation. In addition, you can generate significant cost savings by finding expressive alternatives and packages that you can award to carriers so that they improve their utilization factors. That's very often an entry point for us with large enterprises. And we deal with almost 100 of the world's largest 500 shippers.
The other major challenge for logistics sourcing teams, particularly those in ocean freight, is that as soon as you have agreed on your contracted rates with carriers, things change, your network requirements change. You have new origins or destination ports.
"No matter how big or complicated your negotiation is, we can usually help knock a lot of time off that negotiation"
Every day or every week, you're going back out to market with mini tenders to get updated contracted rates for new lanes. And we have Sourcing Bots that can automate this process so that a Bot for a, let's say, a Nestle, it could be very different from a Sourcing Bot for a Microsoft or a Schneider Electric.
We provide great flexibility in terms of the negotiation process you can automate because underlying Sourcing Optimizer is a very powerful and flexible workflow engine, and we have a native automation framework that will allow you great flexibility in terms of what you can automate.
JM: With ocean freight, requirements constantly change just through the nature of what it is. What technology can you deploy to enable the logistics buyer to get the best spot pricing on the market?
AH: That capability is embedded within the Sourcing Bot so that when requests come in, if there is predictability and regularity in the demand pattern, then that's gathered at the request stage, and that is shared with the carriers. So invariably, their bidding behavior changes if they have that predictability and regularity in transportation requirements.
But it's the same Sourcing Bot that can run your bid event for that predictable demand versus unpredictable demand. And you do get higher rates if there is less committed volume or greater unpredictability in demand.
JM: With the Sourcing Bot you deploy, what's the main win from the buyer's perspective?
AH: It depends on what industry you're in. The primary benefit for some of our customers is the time saved, and I would say that's the number one thing.
One of our customers benchmark the time saving as 93% of the time usually required for these mini tenders in ocean freight or air freight; 93% of the workload was reduced so that they could spend their time on other things. There's many things machines can't do, and there's plenty of other work for people in procurement. So that was seen as the top benefit.
Other benefits included consistency in the process – the process would always follow a predetermined procedure that gathered all of the mandatory data required for successful conclusion and the addition of new rates to the rate card.
Compliance is another benefit and the savings that you get from being able to go to more carriers in parallel. For a Sourcing Bot, it's not a heavy lift to go out to more carriers in parallel. So it's easier to scale a strategy where you speak to many more potential suppliers and generate more competitive tension in the process.
It's also faster to get things done: when the work comes in, it's kicked off immediately. There's no waiting time for somebody in procurement to kick off a mini-tender process.
JM: When that pricing comes back from the Bot, is that typically in a spreadsheet form that's then uploaded into Keelvar?
AH: All of the bid data is collected online; it's not in a spreadsheet. In strategic sourcing negotiations, that could be for 1000s of lanes, and carriers often like to download those and work on a spreadsheet offline and then upload them.
But in faster pace mini-tenders where there might only be a few lanes and a weekly bid, it's all done online. When the Bot has concluded the process, there's a ChatBot interface, where it converses with a human approver to ask, "here are the scenarios we've evaluated, this is the lowest cost option, this is the fastest option," or "this may be the most environmentally friendly option, would you like an alternative to any of these three?" If you wanted something in between A and B, it could produce a scenario between A and B.
You choose to award, and then messages go out to the relevant suppliers – the award notices or notices of not being awarded business. The rates are then automatically added to a rate card so that for new searches for those contracted rates, you will have your known carrier and the known agreed rate.
"Best performing procurement and supply chain management teams are looking at processes to scale that type of sourcing excellence – and automate it"
JM: Does someone that's using this need to have a pretty good grasp of logistics procurement as a category?
AH: Usually the same individuals who operate the strategic sourcing event on an annual basis are the same personnel who are acting as approvers. They know who the strategic relationships are with.
If you wanted to bias in favour of certain providers over others because you have aggregate volume commitments, etc, usually there's somebody knowledgeable who's approving the suggested actions from the Bot.
JM: The shortest route is pretty easy because Google Maps can tell you that. But something like sustainability is a bit more nuanced, isn't it, in terms of how you measure it?
AH: That's correct. But we see a quickly emerging future whereby the Bot becomes more tightly integrated with more data sources. Say you're receiving bids from carriers and it's for transportation from Miami to Oslo.
Suppose you see that there's three carriers competing, and a Bot can interact with third-party data sources to ascertain which vessels will be used, what age the vessels are, and what's the average co2 emissions per TEU on that vessel. In that case, you can start to be very quantitative about expected Scope 3 emissions.
Because you can be quantitative and measure these quite accurately, you can bias. So if you have an internal cost of carbon, for example, then you have all the information you might need for the Bot to automatically evaluate those trade-offs.
I think the best performing procurement and supply chain management teams are looking at processes to scale that type of sourcing excellence and automate it.
Thank you to James Meads at ProcurementSoftware.site