Kinaxis’ Maestro is billed as an AI orchestration platform that revolutionises how supply chain leaders handle and use their data. Built upon three fundamental principles – supply chain data fabric, an intelligence engine, and the user experience – it serves to ease the challenge of gleaning actionable insights from broad data sets, as well as automating processes that are reliant on understanding shifts in that data.
Through AI, it’s a system that users can speak with: ask Maestro a question about your data, and it will give you an answer in real-time. The AI-powered system can also simulate an endless array of scenarios, massively enhancing supply chain leaders’ capacity to prepare for the future against a backdrop of regular and often-decisive volatility around the world. Keen to learn more about the ways in which the firm is leveraging agentic AI in both Maestro and beyond, SupplyChain Strategy sat down with Kinaxis’ Chief Product Officer, Andrew Bell, backstage at Kinexions 2025, to learn more.
The three AI disciplines
Before we get into the finer details, it’s important to understand what agentic AI is and where it sits in the growing family of AI-powered technologies poised to reshape the world. “For supply chain, our view is that there are three AI disciplines that are highly relevant to what we do,” explains Bell, fresh from delivering a fascinating keynote speech to the assembled global supply chain leaders gathered in Austin, on agentic AI. “The first was predictive AI with machine learning, the second, more recently, was generative AI. Continuing on from there would be agentic and autonomous AI.
“It’s not about any one of those on their own,” Bell continues, “but rather how they come together to deliver. When I think about agentic AI, it comes down to what we demonstrated in conference: the ability to chat with your data, to ask questions about your data, to get it presented to you however you want, all based on simple prompts. It’s actually a fusion of generative and agentic AI. There’s the agent that we built that works autonomously based on prompts from users; prompts that are then interpreted by the generative side.”
According to Bell, when it comes to agentic AI, the real differentiator is the notion that it operates on its own, that it operates autonomously as a result of a user prompt or data change conditions. “The idea is that it’s able to make its own decisions as it progresses through a problem; that’s what I find so powerful about it,” he enthuses. “That’s how it differentiates from other forms of automation.”
The democratisation of data
While concerns abound regarding the disruption AI could bring to workforces, namely in headcounts and the nature of their work, Bell stresses that this form of AI, as with the others, is at its best as an enabler rather than replacer. “The first thing to say is that AI on its own, especially in the supply chain space, is not going to solve our problems,” he explains. “It’s not going to deliver the value. Its real value is its democratisation of data access through the combination of the data with tools that have the ability to access and use that data, with AI sitting on top. Then I can get to my data more easily and more quickly, and so can anyone else approved to use the system.
“Users don’t need to learn a system, they don’t need to know how to navigate complex worksheets, set up filters and all the things you do in a traditional context. It means anybody, whether that’s an entry-level planner or a C-level executive can ask data-based questions, run a scenario or a simulation or execute something with less friction. I see it as a democratisation of the power of data and as an accelerant.”
That sense of democratisation extends beyond Kinaxis’ internal use and development of its agentic AI systems, with customers and partners joining the fold to inspire new and iterative action. “We’ve approached it by building an agentic framework first, and that allows for the creation of agents and the running and execution of agents,” Bell elaborates. “That’s step one. Now we’re building our own out-of-the-box agents on that framework, as well as opening that framework up to our customers so they can build their own agents. Customers know their business best, and there might be use cases that they want to apply an agent to that we haven’t thought of yet. They’ll now have the ability to do that.
“From there, we’re using our customers and the challenges they share with us to figure out what we can build or iterate upon next. We’ve started with the ‘chat with data’ agent. Because that was the number one thing: get me access to my data. The next thing is the ability to evaluate two options and execute a change. Merck, who we’re working with, shared an agent that essentially detects late supply and takes corrective action.”
Bell is evangelical regarding the adaptability of its AI framework, allowing agents to be used in isolation, or strung together. “It’s purely going to be based on the natural language prompt from the customer,” he reveals. “The framework will know all the different agents I have access to and so it can either do what the user is asking with those agents or suggest a combination of those agents.”
Data is the key
Data is the crux that all AI roads lead to and stem from. Without high-quality data, AI isn’t capable of delivering on its potential. Creating robust frameworks, exercising high levels of data hygiene, and structuring data stores in an AI-ready fashion are paramount in both the development of agentic AI and the application of those tools. For both developers and users, Bell stresses the fundamental importance of getting that data piece right. He notes, too, that its applicable advice no matter where individuals and organisations are in their AI journey. “There is the ability to start from any position on that journey,” says Bell. “It doesn’t have to be a big bang or a one-size-fits-all. No matter what, though, it is about the data. The agents, the automation, whatever it might be, is only going to be as good as the data that it can access.
“Step one is to understand the problems you’re looking to solve and figure out which data that system would need. We have capabilities that simply do exception reporting where you can implement predefined automations where your team has said ‘these are some processes that we execute on a regular basis, and we have the data, so automate it’. You can then move up the journey and say, ‘No, we’re ready to implement agents and we’re going to start using some proven native ones before going all the way to making our own.’’
“The good news is that some of the foundational requirements apply no matter where you start in the journey. Getting the data and having the right tools in place are going to benefit you across the whole journey. From Covid to more recent impediments to worldwide networks via trade war escalation, significant global interruptions and bottlenecks over the past several years have put enormous pressure on supply chains to adapt at pace. As far as disruptive influences go, agentic AI represents a welcome boon for those who can effectively wield its potential.”
“At Kinexions 2025, we had a presentation from ExxonMobil that noted how people typically think about disruptions as a negative thing, but our job is to build a supply chain that excels at managing those disruptions,” says Bell. “When we do, we have a competitive advantage. Our job at Kinaxis is to provide the tools, systems and capabilities to deliver that competitive advantage to our customers. Disruptions are going to occur. That’s a given. We don’t know what they might be, but they’re going to happen. If we’ve given you the ability to manage them effectively, that’s going to give you a strong competitive advantage.”
- Digital Supply Chain
- Events
- Together in Events