Becoming a data-driven organization isn't some pie-in-the-sky proposition at Monsanto Co. The St. Louis-based agribusiness has formed a center of excellence designed to ensure its business decisions are based on -- and backed by -- data. The "decision science" center of excellence, or CoE, is focused on meshing engineering, data and domain expertise into business decision-making processes, and it uses analytical models and emerging technologies to do it.
In this SearchCIO video interview recorded at the recent MIT Sloan CIO Symposium in Cambridge, Mass., Monsanto CIO Jim Swanson talked about the decision science CoE, investments he's making in emerging technologies and, specifically, how he's tying artificial intelligence (AI) investments to business outcomes as a way of measuring their ROI.
How does AI figure into your IT operations?
Jim Swanson: In my organization, we run what we call a decision science center of excellence. So, I have core mathematicians and statisticians who develop models -- predictive, prescriptive, machine learning and deep learning models. But we also run the center of excellence in the company. We have kind of a hybrid model, where we embed decision science in our different parts of the organization.
But the CoE is there to collect the portfolio, to manage the talent, to orchestrate the work. And so we build the models within the IT organization, and we orchestrate work across the company. But there are also really good decision science developers who are in each part of our business, so we couple the engineering, the domain and decision science across the whole company.
What kind of AI investments are you making?
Swanson: We're focusing on a number of different areas. We have everything from operational research and how we use machine learning for that, to what we're doing with AI and deep learning. So, a good example is in our research area: We're trying to predict via genotype what's going to produce the best phenotype -- a healthy root structure or good leaf structure that can take as much water or nutrients from the soil.
[We use] machine learning and models to, in silico, predict what are the best genes that will become seeds that we put in the ground, and then we produce those in the R&D space. We've been doing that across all aspects of our business, and they have really taken root -- no pun intended -- and drive decisions across the whole company.
How are you measuring the ROI on those investments?
Swanson: We are tying it to real business outcomes. For us, it could be improved net present value of our pipeline. In our supply chain, it would be the cost of goods reduction or efficiencies. In our commercial space, it will be improved net promoter score or revenue lift. We tie [AI investments] to business returns.
And we use a bunch of technologies that are emerging that allow us to accelerate [the ROI] -- whether that be GPU processors to accelerate time to process, working in big platforms like TensorFlow and Google, or using a lot of open source to share models and information. So, we use a number of different technologies. We take a number of different approaches. But we absolutely tie [AI investments] to business return and business value, and that's what gets the organization really excited.