As GVP, Technology and Engineering at Albertsons, Maria Latushkin is transforming how one of America’s largest retailers bridges physical and digital experiences. And her work focuses on scalable solutions that can grow across thousands of locations and hundreds of thousands of associates. Idaho-based Albertsons, with more than 2,200 locations, is the second-largest grocer in North America, after Kroger.
We spoke with Latushkin about the delicate balance of startup-enterprise partnerships; the practical realities of implementing GenAI; and her vision for improving retail experiences with IoT and computer vision. Her insights reveal how large organizations can drive innovation while maintaining operational stability.
This conversation is part of our Early Adopters series highlighting business leaders who are driving digital and AI transformation at major companies. For initiatives to succeed, there must be strong partnerships between Innovation teams and business executives. Through these interviews, we share perspectives from functional leaders who are putting emerging technology into practice.
• • •
Startup Partnerships
What characterizes your most successful startup relationships?
I view startup-enterprise relationships as long-term partnerships focused on co-creation. The most successful startups solve our immediate business problem first, and then extend the partnership and evolve the solution against our roadmap.
These partnerships succeed when both parties benefit. As an enterprise, we gain from evolving solutions that grow with our needs. In turn, we help the startup develop a stronger product and business value proposition.
The startups I admire most take this further. They’re able to solve a business problem, dissect it, and then reposition their core solution for different use cases or industries. This platform approach allows their solutions to evolve beyond single use cases into broader value propositions.
The key is finding the right balance. The startup solution must be specific enough to solve our needs while remaining flexible enough to serve other customers, which ensures long-term sustainability for both parties.
In terms of startup relationships that don’t meet expectations, what are the common pitfalls you’ve observed?
Overselling capabilities is a major pitfall. Many startups fall into this trap trying to put their best foot forward. As a large enterprise, we understand a young startup won’t have everything ready at first. We’re looking for a partnership, a relationship, a roadmap. But when startups claim to have everything ready, it often breaks down during rollout, whether from technical challenges or operational support issues.
Product integration is another major watch-out for startups. For a large enterprise, if a startup’s solution can’t fit into our existing systems, implementation becomes nearly impossible. When a solution requires a greenfield approach, it rarely succeeds. Buyers will rarely agree to completely overhaul their existing systems. The solution needs both the right product architecture and business architecture to wedge itself surgically into our ecosystem and grow over time.
A startup solution needs both the right product architecture and business architecture to wedge itself surgically into our ecosystem and grow over time.
Pricing can create significant challenges, especially early on. Often, I see attractive pilot pricing that doesn’t translate to full deployment. For example, if a solution is location-specific, we have to ask: will it still be cost-effective when multiplied across thousands of locations? Similarly, with seat-specific pricing, we need to consider the cost impact across hundreds of thousands of associates. Can we maintain positive ROI at that scale? The key is finding a cost model that’s both relevant and sustainable for both parties.
When you decide to evaluate and implement a startup solution, are there any internal stakeholders that are crucial to involve on your side?
I always advise startups to ask about key decision-makers during their pitches. Decision-makers vary at every company based on organizational structure and investment size.
Often startups tell me: “I thought you wanted to move forward with installation?” But when I dig deeper into what led to this assumption, it turns out they were talking to people at the wrong level and/or the conversation was theoretical. Having someone like your product isn’t a straight line to having a budget and securing all necessary approvals.
At the same time, you do need to ensure the end users become your advocates. Whether that’s the business end user or a developer in the case of technical products, these are the people who can best explain the business problem and the solution’s value. End users champion your solution, driving both initial adoption and ongoing success.
Do you currently engage with startups as an advisor or an investor?
I’m not an investor, but I do serve as an advisor. Having worked at both startups and enterprises, I understand how challenging it can be for startups to break into large companies. The advice I often give is: don’t focus on landing the most prestigious customer. Focus on finding the right customer for your stage.
I saw this firsthand at some large companies I worked at in the past, where our needs were so specific and complex that it was often overwhelming for startups to fulfill them. In some cases, it wasn’t even financially viable for a startup to take us on as a customer.
From my startup experience, I know how tempting it is to pursue major clients. But first, you must honestly assess if you can support them sustainably — both from a technical and operational standpoint.
GenAI
How has GenAI transformed your organization? Can you share an example of a quick win?
One area where GenAI is already showing strong impact is in development and coding assistance. AI tools act as a great equalizer across our large, diverse, global workforce. For our junior team members, they reduce ramp-up time on different codebases and skill sets, giving them an extra boost in productivity. For senior developers, they eliminate menial tasks. While the benefit varies by experience level, it’s equally valuable across the board.
GenAI has also improved information processing and retrieval for companies — spanning both customer-facing chatbots for service and support, and internal systems that help associates search through documentation and find answers to their questions. Companies are seeing significant benefits in both our external customer service and internal knowledge management.
In what areas, broadly speaking, have you found GenAI to be less beneficial?
At this moment in time, and I think this will change, GenAI is less beneficial when trying to transform highly complex and bespoke processes. Think of these processes as links in a chain: you might successfully apply AI to one link, but then hit a barrier at the next — whether from human dependencies, legacy system limitations, or other technical constraints.
In the beginning, many organizations approached GenAI by asking, “What are our biggest problems?” The assumption was that solving the biggest problems would yield the biggest ROI. But these big problems are often very complex processes. To transform them, you have to really dig into the complexities, deal with different systems, and handle various inputs. While this analysis often leads to valuable process improvements, the amount of foundational work needed before GenAI can even help sometimes makes these projects less attractive.
Have there been any tactics that you can share that have proven to be successful in driving GenAI adoption among your teams?
There’s no secret formula, but we’ve found a few approaches that consistently work. First, you need to find use cases that demonstrate clear success. People need to see actual benefits. Second, education is crucial. We’ve over-invested in education, looking at different personas and tailoring our messaging for each group.
A tailored approach is essential because different teams have different needs. Developers need technical deep-dives, while business teams need clear explanations of how GenAI can help their day-to-day work. As opposed to lengthy training courses, we focus on bite-sized education that fits into people’s existing workflows while still transforming how they think about their work.
For our technical teams, we identify and develop internal champions — enthusiastic team members who receive intensive training and then build communities of practice to share their knowledge. This train-the-trainer approach helps spread both expertise and excitement. But enthusiasm alone isn’t enough; we rigorously measure results to ensure we’re delivering real business impact, not just running a hobby project.
Tech beyond GenAI
Beyond GenAI, what other technology trends are you watching closely?
Everybody is really focused on GenAI, but there’s so much happening in the broader AI ecosystem that shouldn’t be overlooked. The fundamental shift is that the world is moving much faster, with more connectivity and real-time decision-making required of enterprises.
We’re also processing information differently now; it’s becoming much more visual and physical. IoT and computer vision are becoming critical technologies, enabling enterprises to gather data from devices, extract insights, and create new customer experiences. These real-time feedback loops are transforming how businesses connect with and understand information.
When customers visit our stores, we want them to benefit from both physical interactions and the rich information available online.
The goal is not to completely change how we process information, but to augment and enhance our existing capabilities. For example, in our stores, we’re working to create seamless transitions between online and physical experiences. When customers visit our stores, we want them to benefit from both physical interactions and the rich information available online. This might mean helping them move more effectively through aisles, learn about products, understand which products go together, or even start their shopping journeys before they arrive at the physical store.
How are you thinking about the integration of IoT and smart devices in stores, particularly for inventory management and customer experience?
Retailers are striving to create the most relevant and useful customer experience. This means removing friction and helping customers move more effectively through aisles, find the products they’re interested in, and learn about those products. We want customers to understand which products go together and even start their shopping journeys before they arrive in the physical store.
IoT and computer vision are incredibly important for creating a more integrated, intelligent retail experience that bridges digital and physical spaces.
As you integrate the online and in-store experiences, how do you approach modernizing your legacy systems?
Our approach is like peeling an onion. We can’t create disruption for our business, so we try to surgically remove certain pieces of the legacy systems where possible. Sometimes you have to bite the bullet, especially when dealing with a monolith that cannot be pulled apart. Mostly, we strategically integrate new technologies into existing systems, modernizing piece by piece to reduce any disruption to the business.
Paulina Karpis leads Early-Stage Platform for B Capital, a global multistage venture firm investing in B2B startups.
Photo courtesy of Albertsons Companies
