According to Forbes, Mars has invested over a billion dollars in its digital transformation and is now “extremely bullish” on AI. Shubham Mehrish, the company’s global head of Generative AI, revealed they’re using AI for marketing, science, and diagnostics, including a veterinary radiology tool called RapidRead trained on 16 million images. On the consumer side, they’ve launched initiatives like the Greenies Canine Dental Check and IAMS Poopscan. Internally, their AI platform, Mars’ MAX, gained 20,000 organic users in months, and the company established a Responsible AI Governance board 2.5 years ago to oversee all use cases.
The Scale Problem
Here’s the thing that jumps out. Mars isn’t just one company. It’s a sprawling conglomerate of fundamentally different businesses. You’ve got the classic CPG side pumping out M&M’s and Pringles. Then you’ve got a massive network of veterinary hospitals and a diagnostics arm. Applying AI effectively in a factory optimizing Snickers production is a world apart from using it to read a dog’s X-ray with 95% accuracy.
Mehrish admits this is what keeps him up at night. Building a platform that’s actually scalable across all that? It’s a monumental IT and operational challenge. You can’t just buy an OpenAI API key and call it a day. The “foundations” he mentions—data cleanliness, integration, change management—are the unsexy, brutally hard parts of corporate AI. And that’s before you even get to the culture shift of 170,000 “Associates” working alongside AI agents.
Consumer-Facing: Gimmicks or Game Changers?
Let’s talk about those consumer apps. Scanning your dog’s teeth or its poop with your phone is certainly attention-grabbing. It’s a clever way to use AI for direct customer engagement and data collection in the pet care space, which is a high-margin, emotionally-driven business for Mars.
But I’m skeptical about the long-term value. Is Poopscan a fun novelty or a genuine health tool that creates lasting brand loyalty? The risk is that these become marketing-led experiments that don’t fundamentally transform the core service. The real potential, which they hint at, is in the veterinary diagnostics like RapidRead. That’s where AI can directly impact efficiency, accuracy, and cost in a professional setting—the stuff that moves the needle on the bottom line.
The Responsible AI Dilemma
It’s encouraging that Mars started its Responsible AI governance 2.5 years ago, driven from the bottom up by employees asking ethical questions. That’s often how it should work. Having a cross-functional board and a “human in the loop” model is table stakes for any major corporation now, especially one dealing with animal health data.
But the proof is in the pudding. “Every use case goes through the framework” and “each prompt goes through guidelines” sounds great in a Forbes interview. The reality in a decentralized, multi-billion dollar empire is messier. Enforcing those guardrails uniformly, from a marketing team generating ad copy to a vet interpreting an AI-assisted diagnosis, is a relentless task. Their early start gives them an advantage, but the complexity of their own organization is their biggest test.
Betting on a Future They Can’t Fully See
The most fascinating part is Mehrish’s vision of the future. He’s talking about AGI and “reimagining” the roles of their people. That’s not just about efficiency; it’s about existential preparation. When the head of Gen AI at a candy and pet food giant is pondering superintelligence, you know the conversation has shifted.
Mars’s bet seems to be: we missed the early digital wave, but we won’t miss the AI wave. They’re throwing money and platform-building at the problem. But history is littered with corporate “digital transformations” that fizzled. The difference now is the palpable pressure. If they don’t figure out how to scale AI across those factories, hospitals, and brands, someone else will. The question isn’t if AI will change packaged goods and pet care. It’s whether a 113-year-old family-owned behemoth can bend it to its will before it gets disrupted.
