Norwegian salmon-farming firm Mowi recently announced an expanded partnership with Helsinki, Finland-based AI platform RELEX Solutions to digitize and forecast raw material planning operations.
The companies will work in tandem to provide more visibility into market supply and demands, improve forecasting accuracy, and produce additional profit data using AI to condense spreadsheet-based processes, business intelligence tools, and enterprise resource planning (ERP) workflows into a single database.
“We are proud to expand our partnership with Mowi,” RELEX Solutions Group CEO and Co-Founder Mikko Kärkkäinen said in a release. “Planning for fresh, highly perishable products such as salmon requires precision, adaptability, and close coordination across the value chain. By connecting demand and raw material planning in one environment, Mowi will be able to make faster, better-informed decisions that support profitability and reduce waste.”
RELEX Vice President of Industry, CPG, and Manufacturing Rohit Tripathi told SeafoodSource he hopes deals like this will help the seafood industry look beyond the buzzwords of AI and focus on the capabilities digitizing processes can provide to gain better insight. His career includes nearly 30 years of work experience in supply chain optimization software at companies like SAP and Boston Consulting Group.
"I think the one thing that I would love the seafood industry to take away from this whole AI [movement] with planning is, I would ask them to focus on AI that delivers them the [returns on investment] today, not the future," Tripathi said. "It should be AI that scales and does not become a pilot project, and it should be the AI that helps their planners grow and be critical and become strategic thinkers, not replace them."
Tripathi added that his work has taken him to several different food production facilities, and he's met with planners who spend approximately 60 percent of their time "cutting and pasting information from one Excel sheet, putting it in another Excel sheet, and formatting it."
"[I'd rather see planners make] strategic decisions because they know that business," Tripathi said. "Then, let AI do all that [data collection and formatting]."
Mowi and RELEX Solutions began working together a couple of years ago. Tripathi said the company wanted to digitize the value chain from the net pen to the customer with AI-driven demand forecasts. Once the software proved valuable, Tripathi said the two agreed to expand the partnership companywide. Now, Mowi is one of RELEX Solutions’ key customers and its first pure seafood client.
"I think what is happening is as the size and scale of these operations is growing, certainly the executives are now asking how [they] can make things better," Tripathi said. "The more interesting part with fish is, and this is the key supply chain problem that RELEX solves very well, it's known as an inverse bill of material. Almost all of the supply chains, the way it works is you get many inputs, you assemble them together, and then you create one output or a product. With fish, you start with one input and make many things from it. You start with one carcass, or one cow, or one pig, and then you make many things out of it; the supply chain problem is flipped."
RELEX Solutions' software can break data down into supply components and demand components.
On the supply side, RELEX, for example, is able to work with capacity bottlenecks to look at individual packaging and capacity lines. RELEX also inputs data on shelf life, with fresh fish needing forecasts in days not months.
"You have to address everything in that window," Tripathi said. "That's why if you talk to a fish executive and you say, 'Hey, let's talk about six to nine to 12 months of supply chain planning.’ They're like, 'Yeah, that's a good idea, but all I really care about is one month and two months.' RELEX helps them address that."
On the demand side, Tripathi said the software works with historical sales data, which is inputted into the machine learning system and developed into a customized algorithm. Most importantly, Tripathi said data privacy is of the utmost importance, and a company's historical sales data will never be made public. Once data is inputted, the algorithm is able to isolate and understand where both sales spikes and shortfalls occurred. It then correlates that data with any possible linkages like weather events or promotional sales events.
Then, Tripathi said, the data on both ends can be used to make strategic decisions based on market needs – something already being done by planners but now at a faster, more in-depth analytical level.
"For example, in the next few months this year, we know there is [the possibility of] a Super El Niño," Tripathi said. "Some regions will have very warm temperatures, some regions will have colder temperatures, and then there'll be a lot more rain in some. That for sure is going to impact the production of fish, like fish farms, so now that's a scenario that you can run."
As with any new technology, Tripathi said there will still be errors, but he said what sets RELEX apart is any planning deployment requires a "human loop" to correct or approve measures before implementation. Eight times out of 10, the model predicts correctly, yet he still sees resistance from planners when previewing the software. Trust is built through a system called "touchless planning" that allows the planner to go in and change the algorithm or reset guardrails for measurement.
"You require that precision, adaptability, and close coordination across the value chain," Tripathi said. "By connecting demand, raw material planning, and your constraints all in one platform, it helps because everybody is working on the same data points and assumptions, using AI that is explainable. It does not hallucinate, and it always has the human planner in the loop."
