Escorts Kubota enlists AI to reinvent railway, construction, and agriculture

“When you start talking about AI, ML stuff, it touches almost every department of the organization and every user of the organization,” he says.

Given this broad scope, it would be easy to fall into the trap of adopting ad hoc projects across the business. Instead, Kakkar has created pillars each project must fall into: customer service, predictive maintenance, supply chain digitization, and personalized marketing.

Kubota has projects across these pillars in various stages of maturity, with some already live and some still in experimentation. Kakkar’s litmus test for pursuing a project depends on whether it has a clear purpose, goal, and measurable objectives. If all three are in place and there is visibility at the board level, Kakkar says the project will be readily funded.

AI in action

One such project that Escorts Kubota has pursued concerns personalized marketing. The company segments customers who visit its website into various profiles and then uses AI and ML to provide tailored agronomic advisory.

For example, a customer who peruses the offers on their website will primarily get communications around discount schemes. A customer who looks at different content, such as those relating to horsepower or implements, would be served different content.

“Then we are giving the USP [unique selling proposition] on how our products are suitable for their respective crops, what kind of horsepower they can use, and what kind of implement they should use to increase the yield of their crop,” says Kakkar.

This initiative currently uses static data, but Kakkar has bolder ambitions for its agronomic advisory. He says that it may start ingesting online data through IoT devices so that it can provide recommendations based on the actual conditions of its feed.

“So whether he needs to use pesticide, what could be the percentage, what is the soil condition, how they could use that, and what is their tractor uptime — how they can ensure the tractor is up and running during the peak season of their harvesting or maybe sowing season,” he says.

Kakkar also believes that AI, ML, and automation projects should be inclusive. “You need to ensure everybody understands what you are trying to do and be part of your project. Whether the role is small or big — they should be part of the project,” he says.

Escorts Kubota’s ecosystem of stakeholders includes the organizations they sell to and, just as critically, the end users of these products. Some stakeholders may not have a tech background and thus not immediately appreciate the need for cutting-edge technologies.

Kakkar structures its communication around two pillars. The first is based on how the solution will help them. For example, Kakkar says that they might share how a tool would free up time for higher-level analysis rather than losing time to routine, day-to-day operations.

The second involves mapping their business or career goals. “If someone would want them to be elevated to the next role, we tell them, ‘If this kind of model would help you to manage your job, then you could focus on the bigger role. You can take more responsibilities,’” he says.

Such messaging has its challenges. If a technology is good enough to fulfill a task, what stops it from taking over the rest of their work? “So they also have a fear of losing the job,” says Kakkar, concerns which are addressed through discussions with management and HR where they clarify that tech-enablement is the only goal, not workforce reduction.

Once they are successful, Escorts Kubota can assist these stakeholders in adopting their solutions. For example, some of their potential customers use traditional digital dashboards as the legacy solution.

Kakkar says they are keeping this underlying technology the same.

“We are accumulating the data by creating the pipeline from all the sources, putting it into the data lake, and helping them to understand the unknown in the data, which they are not able to visualize by having the static database, static KPIs, or dashboards,” he says.

He points to data cleanliness as a major challenge in this workflow. “Because when you are ingesting the data from your legacy system, you need to ensure the integrity of the data,” Kakkar says.

A related challenge is that data must be labeled and defined consistently. Kakkar says that they created complete mapping access for everyone’s reference.

“We had created two-dimension mapping tables: source, target; so that the people would know, ‘Okay, if this is the source, how the data was coming, this is the target on the other side, and now how this has been converted,’” he says.

Although Escorts Kubota is still early in its AI, ML, and automation journey, Kakkar acknowledges that the human element will play the most crucial role. Because people are often hardwired to the status quo, Kakkar says that any project rightfully involves changing the mindset of the people involved.

“We need to do a little more homework, help them to visualize the technology, and how that can help them. And also how they could improve their business rather than just telling them, ‘This is the technology project they need to adopt,’” Kakkar says.