Artificial Intelligence and the Supply Chain in a World of Converging Agribusiness Software
This week, I read news from Agriculture Dive on an AI chatbot that can help predict farm supply chain disruptions.
It caught my eye, specifically if we extrapolate the functionality more upstream in the agriculture value chain:
Helios on Thursday launched Cersi, a chatbot that harnesses billions of climate, economic and political signals to forecast potential supply chain risk down to the farm level. The virtual assistant is part of Helios’ larger risk management platform, which was launched in December, and is meant to help large food companies get a better handle over their agriculture supply chains.
Helios focuses more on the downstream supply chain for CPG companies securing grain supply from farms.
I think this presents a valuable example of two things directly related to upstream portions of the value chain:
Opportunity for product evolution in upstream supply chains.
An example of where a Large Language Model ChatBot interface isn’t the best default approach.
I also believe this functionality will become a future feature contributing to digital battlegrounds in the agribusiness software.
Opportunity for software evolution in upstream supply chains.
Every year, there are supply chain challenges in getting input products where they need to be, when they need to be there, leading to missed opportunities for input manufacturers and retailers, increasing inventory levels, and logistics costs— especially in the current higher interest rate environment.
I am not suggesting we need standalone AI-enabled logistics software. I am suggesting the opportunity to embed in agricultural software (ag-specific vertical SaaS) smart logistics capabilities into ERPs, systems of action, crop planning tools, and other software that connects demand signals with supply realities. Taking into account weather, crop forecasts, insect forecasts, previous years’ sales, commodity prices, and more to deliver suggestions in real-time, it can benefit farmers, retailers, and input manufacturers.
Suppose it costs an integrated retailer/distributor (Simplot, for example) 2% of sales to store and distribute $3 billion of input products annually, and they can improve that number by 5% and get it down from 2% to 1.9%. In that case, that’s $3 million in annual savings in distribution costs, not accounting for return on capital tied up in inventory, incremental sales, or program optimization they may take advantage of.
This can be extrapolated to input manufacturers, standalone distributors, or retailers.
The critical aspect is effectively embedding this capability directly into agronomists’ and procurement managers' software— not having a separate place.
That leads us to point #2: the mechanism to deliver those AI insights.