Building AI and Web3 Products: Lessons from a Minnesota Product Manager
It’s easy to assume that serious product work—especially in AI and emerging technologies—happens exclusively in Silicon Valley, New York, or maybe Austin. I get that assumption. But I’ve spent years building AI and data products from Minnesota, and I can tell you with confidence: some of the best product thinking happens far from the hype centers. In fact, Minnesota’s particular culture and remote-first tech landscape have shaped how I approach product management in ways that have made my work stronger.
This isn’t a post about why Minnesota is “the next Silicon Valley” or any such nonsense. It’s about what actually works when you’re building complex, data-intensive products from the Upper Midwest—and why Minnesota’s pragmatism, collaborative spirit, and growing distributed tech talent pool make it a surprisingly excellent place to ship serious AI products.
The Reality of Building AI Products in 2025-2026
Let’s start with what’s changed. When I started in product management, “AI” was still a fuzzy concept for most stakeholders. Today, it’s embedded in nearly every product conversation. But that abundance of interest masks a hard truth: building AI products is exponentially more complex than building traditional software.
The core challenge isn’t the AI itself—it’s the data, the infrastructure, the governance, and the human behavior that surrounds the AI. At Veeam, I work on data management and backup solutions that increasingly incorporate machine learning and AI capabilities. The technical feasibility of AI is rarely the bottleneck anymore. What matters is:
- Data quality and provenance—You can’t ship AI products without clean, well-understood data pipelines. This is boring work, but it’s foundational.
- Regulatory and compliance requirements—AI introduces new compliance questions around bias, transparency, and data usage that vary by geography and industry.
- User trust and explainability—People don’t want a black box. They want to understand why the system made a decision, especially in enterprise contexts where billions of dollars of data are at stake.
- Infrastructure and cost management—Running AI models at scale is expensive. A PM has to understand the computational trade-offs of different approaches.
- The team’s capability—You need data scientists, ML engineers, platform engineers, and product folks who can all communicate across disciplines.
From Minnesota, without the constant influx of venture capital and the pressure of startup culture, I’ve found that these challenges become clearer. There’s less tendency to oversell what the technology can do and more focus on what customers actually need.
Why Minnesota’s Culture Shapes Better Product Decisions
Minnesota has a reputation for being pragmatic, collaborative, and straightforward. People joke about “Minnesota Nice,” but there’s something real underneath that stereotype when it comes to product work.
The Pragmatism Principle
Minnesota business culture values practical results over hype. This isn’t unique to Minnesota, but it’s pronounced here. In my experience, this pragmatism translates directly into product decisions:
When I’m evaluating whether to invest in an AI feature, I’m less likely to be swayed by “because AI is the future” and more likely to ask: “Will this actually solve a customer problem? Can we explain it? Can we support it? What’s the ROI?” In Silicon Valley, sometimes the narrative comes first and the customer value comes second. Here, it’s the reverse.
I’ve seen teams in other regions ship ambitious AI features that looked great in demos but failed in customer hands because they solved problems nobody had. Minnesota’s skepticism acts as a useful filter. We ask harder questions earlier, which means we build fewer features that embarrass us later.
Collaboration Over Competition
Minnesota’s business community has a less cutthroat feel than some tech hubs. People are willing to share learnings, collaborate across companies, and think about the region’s growth as a shared endeavor. This affects how I think about product partnerships and ecosystem decisions.
At Veeam, we work with partners—software vendors, cloud providers, service providers—to build integrated solutions. In Minneapolis-St. Paul, there’s a genuine willingness to collaborate that makes these partnerships easier to negotiate and more productive. People here aren’t as obsessed with extracting maximum leverage from every deal. They care about building something that works.
This collaborative mindset has influenced how I think about API design, data sharing, and partnership architecture. We design products to play well with others, not to lock people in. It’s not just nice; it’s smart long-term thinking.
Long-Term Thinking
Minnesota companies tend to think in decades, not quarters. This is partly cultural—many of our largest employers (Target, 3M, Best Buy, Hormel) were founded here and have stayed here—and partly structural. We’re not as dominated by venture capital, so there’s less pressure for hockey-stick growth and exit events.
For AI products, this long-term horizon is invaluable. Machine learning projects don’t reach full maturity in a quarter. They need time to accumulate data, for models to improve, for customer workflows to adapt, for support teams to build expertise. Minnesota’s patience with longer product development cycles means I can take the time necessary to get AI features right, rather than rushing something half-baked to market.
Building AI Products as a Remote-First Minnesota PM
Here’s something that’s changed dramatically in the last few years: remote work has democratized where good product management can happen.
I manage product from Minnesota, but my team spans the country and the world. This would have been nearly impossible to do effectively before 2020. Today, it’s not just possible—it’s normal. And this shift has specific implications for how I work:
Talent Access Isn’t Geography-Dependent
Ten years ago, if you wanted to build an AI product team, you had to hire in San Francisco, Seattle, or New York. Today, top data scientists, ML engineers, and product designers are distributed globally. I can hire people in Minnesota who are as talented as anyone in Silicon Valley—often at lower cost and with better quality of life.
This has attracted serious tech talent to Minnesota. People are choosing to stay here or move here because they can have fulfilling tech careers without the Bay Area’s cost of living or the constant startup frenzy. For product managers, this means access to institutional knowledge and stability. People aren’t jumping ship every 18 months for the next unicorn. They’re building real products with depth and care.
Asynchronous Communication Matters More
When your team is distributed across time zones, you can’t rely on constant synchronous communication. This forces clarity. Slack messages aren’t enough. You need written specifications, documented decisions, and recorded walkthroughs.
For AI products specifically, this is a huge advantage. AI work is complex. When I have to write down why we’re pursuing a particular modeling approach or why we’re making a specific product trade-off, I think more clearly. I catch logical gaps that I might have glossed over in a quick meeting. My team can consume this information on their own time, think deeply, and come back with better feedback.
This is the opposite of the “move fast and break things” ethos, but for AI products, that’s exactly what you want. You can’t iterate your way out of a fundamentally flawed approach to data governance or model explainability.
The Minnesota Advantage in Remote Work
Minnesota has adapted well to remote work culture while maintaining a strong local community. We have multiple tech hubs (Minneapolis, St. Paul, Rochester, Duluth) with co-working spaces, tech meetups, and local tech communities. As a PM, I can do focused remote work but also connect in person with other Minnesota product folks for learning and relationship-building.
This hybrid approach—mostly remote, occasionally in-person—seems to be the sweet spot for serious product work. I get the talent access and flexibility of remote work without losing the serendipitous conversations and deep relationships that come from local connections.
Specific Lessons from Launching AI Products
Over the last few years, I’ve been directly involved in launching AI-powered features and products. Here are the lessons that have stuck with me:
Lesson 1: Data Quality Is Non-Negotiable
This is where I see teams fail most often. They get excited about the machine learning possibilities and build a beautiful model, but the underlying data is biased, incomplete, or poorly documented.
Before you build any AI feature, you need to understand your data: Where does it come from? How representative is it? What biases might exist? What are the gaps? What happens when the real world throws edge cases at your model?
I spend an unusual amount of time—for a PM—in data lineage discussions and data quality audits. This isn’t glamorous work, but it’s where most of the product risk lives. A beautiful algorithm trained on biased data will make biased decisions. No amount of clever UX design fixes that.
Lesson 2: Start with Explainability
In enterprise software (which is where Veeam operates), nobody wants a black box. When an AI system recommends an action—especially something related to critical data—users need to understand why.
I always push for explainability as a first-class feature, not an afterthought. What factors influenced this decision? How confident is the model? What would need to change for a different recommendation? These aren’t nice-to-haves; they’re requirements.
This is easier to build in from the start than to retrofit later. It also forces clearer thinking about the problem you’re solving. If you can’t explain why the AI made a decision, maybe you don’t understand the problem well enough yourself.
Lesson 3: Human-in-the-Loop Is Usually Right
Full automation sounds great in theory. In practice, most AI features work best when there’s a human in the decision loop, at least initially.
I design with the assumption that users will want to review AI-generated recommendations before acting on them. The AI can flag issues, suggest actions, and surface patterns that humans might miss. But the human makes the final call. This isn’t a limitation of current AI—it’s usually the right product model.
Over time, as the system proves itself, you can gradually increase automation. But starting with human oversight builds trust and gives you room to improve the model based on real-world feedback.
Lesson 4: The Cost of Running AI Compounds
AI models have infrastructure costs that scale with usage. This is obvious in theory but easy to ignore in practice.
Early on, when you’re building the feature with small datasets, costs are negligible. But when you scale to thousands of users running models continuously, those costs become real. I’ve learned to model this out early and be explicit with stakeholders about the infrastructure investment required.
Sometimes this means choosing a simpler, less powerful model. Sometimes it means building infrastructure to batch process AI work rather than running it in real-time. Sometimes it means the economics just don’t work for a particular feature idea. These are hard conversations, but they need to happen upfront.
Lesson 5: AI Training and Adoption Takes Longer
When you ship a new AI feature, users often don’t know how to use it effectively. They may distrust it. They may not understand what it’s good for. This requires customer education and support investment that goes beyond typical feature launches.
I budget for this explicitly: training materials, webinars, customer success touchpoints, and support ticket preparation. In my experience, this investment pays back in adoption rates and customer satisfaction. It also surfaces product issues that you might have missed otherwise. Customers using the feature in the wild will find edge cases and workflows you didn’t anticipate.
Minnesota’s Emerging Role in the AI and Data Economy
Minnesota isn’t known as an AI hub, but that’s changing. We have:
- Strong data infrastructure companies—Veeam (my employer) is a global leader in data management and backup. We’re increasingly important as data volumes grow and AI applications proliferate.
- Established software companies investing in AI—Companies like Target, Best Buy, and others based here are building AI capabilities for retail and logistics.
- Healthcare and medical device companies—Minneapolis-St. Paul has a massive healthcare sector with significant AI and data science opportunities.
- A growing startup ecosystem—Companies like Optum (insurance and healthcare), UnitedHealth, and others are attracting AI talent and funding.
For product managers, this creates an interesting opportunity. Minnesota companies are serious about AI and data, but they’re approaching it with practical, customer-focused thinking rather than hype. That’s a competitive advantage.
The Outdoors and Clear Thinking
I want to add something that might seem tangential but actually matters to how I do product work: Minnesota’s outdoors.
I spend a lot of time outside—hiking in the Boundary Waters, running along the Minnesota River, kayaking on our lakes. This isn’t just recreation; it’s part of how I think through hard product problems. There’s something about being in nature that clarifies thinking. When I’m wrestling with a complex product decision, a few hours outside often brings clarity that I couldn’t find in an office.
This might sound like lifestyle branding, but it’s genuine. And I think it’s actually relevant to product work. Some of the best product thinking requires stepping back, clearing your head, and approaching problems fresh. Minnesota’s abundant natural spaces make that easy.
Practical Takeaways for Product Managers Building AI Products
Whether you’re based in Minnesota or elsewhere, here’s what I’ve learned about building AI products successfully:
- Prioritize data quality over algorithm sophistication. A simple algorithm with great data beats a complex algorithm with mediocre data, every time.
- Make explainability a first-class feature. Build it in from day one, not as an afterthought.
- Plan for longer product cycles. AI products take time to mature. Don’t set expectations based on traditional software development timelines.
- Consider the infrastructure costs explicitly. Model out what it costs to run your AI features at scale, and be honest about those costs early.
- Start with human-in-the-loop workflows. This builds trust and gives you runway to improve your models based on real feedback.
- Invest in customer education and training. New AI capabilities require more customer education than typical features.
- Think long-term and locally. Build products that matter to customers in your community. Build a team that can stay together. These compound over time.
- Embrace asynchronous communication and clear documentation. This forces clearer thinking and scales better than constant meetings.
Why Minnesota Matters for the Future of AI Products
We’re at an inflection point with AI. The technology is real and increasingly powerful. But the novelty is wearing off, and the hard work of building sustainable, trustworthy, valuable AI products is just beginning.
This is exactly the kind of work where Minnesota excels. We have the talent, the infrastructure, the large established companies experimenting with AI, and a culture that values pragmatism and long-term thinking. We’re not going to out-hype Silicon Valley, and we don’t need to. We’re going to build better products.
For product managers in Minnesota, this is a remarkable time. The opportunity to build serious AI and data products without leaving the state has never been better. And the talent pool—both technical and product-focused—is deeper than most people realize.
If you’re considering where to build your product career, don’t automatically assume you need to move to the coast. Minnesota’s tech scene is real, growing, and built on a foundation of pragmatic, customer-focused product thinking. That’s not a bad place to build something meaningful.
