The Art of Prioritizing AI Features Amid Uncertainty and Skepticism

As a product manager, your role is to organize ideas into coherent, valuable work — to surface the best opportunities, shape them into actionable features, and bring forward new ideas that drive business outcomes. But when it comes to AI-centric features, this role takes on a unique complexity. Our business partners often come to the table with varied and sometimes conflicting expectations of what “AI” means for the processes and outcomes they own. For some, it’s about automation; for others, it’s about prediction, personalization, or entirely new capabilities. Navigating these expectations requires not just technical understanding but a sharp sense of how to prioritize opportunities that align AI’s potential with real business value. In this post, we’ll explore how proposing and prioritizing AI-driven work is different — and why product managers need a distinct mindset to succeed in this evolving landscape.

I often wish AI features were just like any other — that we could simply measure them against clear user value and impact, prioritize them accordingly, and move forward. But alas, AI doesn’t work that way in most organizations. More often than not, you’ll find yourself presenting an AI-powered solution to a well-known & understood pain point, only to have the conversation spiral into skepticism about the solution itself. People will debate the model’s accuracy, the data quality, or the fairness of the algorithm — and in doing so, they often forget about the original problem you’re trying to solve. As a product manager, part of your job is to help the team stay focused on the pain point and the desired outcome, not just the shiny (or controversial) AI method you’re proposing to get there.

Ideas that involve the application of AI can be especially tricky these days because of this very human dynamic. As a product manager, you have to stand confidently in two places at once: firmly committed to the importance of solving the pain point, and equally confident in the solution you’ve carefully shaped through diligent work across the organization. You’ve assessed the landscape, weighed alternatives, measured potential impact, and arrived at a proposal that deserves leadership’s support to invest and move forward. Yet even then, you’ll encounter a mix of reactions — some stakeholders are enthusiastic, eager to push the boundaries of what’s possible; others are cautious, worried about risk and feasibility; and some are hesitant because they’ve been burned by past AI efforts that overpromised and underdelivered. Your role is to balance these perspectives, keeping the team aligned on why the work matters and why this approach is the right one now.

When you’re presenting your proposed AI approaches — along with the supporting budget needs, resource asks, timelines, and expected impact — it’s crucial that you deeply understand your audience and where they fall across this spectrum of enthusiasm, caution, or hesitation. A one-size-fits-all pitch won’t work. For the enthusiastic, you’ll want to temper excitement with realistic framing so expectations stay grounded. For the cautious, you’ll need to address risks head-on, showing how you’ve thought through potential pitfalls and built in safeguards. And for those who are hesitant or skeptical, it’s often less about the technical details and more about rebuilding trust: demonstrating why this time is different, and why the lessons of past failures have informed a more thoughtful, viable approach. Tailoring your message to meet each audience where they are is just as critical as the solution itself — because no matter how good the idea, if you can’t bring people along, you won’t get the buy-in you need to deliver it.

You’ll likely need to prepare key partners — business stakeholders, security teams, compliance teams — well before you bring your proposal to the wider group for approval. Each of these groups has specific concerns: security will focus on data handling and risks, compliance will want clarity on regulations and ethical use, and business partners will care about alignment with goals and outcomes and begin to consider the organizational change implications. Addressing their needs early, sharing tailored information, and incorporating their feedback upfront helps smooth the path — so when it’s time to ask for broader support, you’re not blindsided by last-minute objections or delays.

Your biggest challenge will often be the knowledge gaps — not just among stakeholders, but across the entire organization. The AI landscape moves so quickly that even experts struggle to stay fully up to date, meaning most people are working with partial or outdated understanding. On top of that, there’s a lot of fear, uncertainty, and doubt (FUD) swirling around AI: concerns about hype, ethics, job impacts, or technical risks. As a product manager, you’ll need to proactively counteract this, providing clear, grounded explanations that cut through the noise and help your team focus on the real opportunities and challenges at hand.

Another layer of complexity comes from the fact that people already experience AI in their daily lives — sometimes in ways they don’t even realize. As I wrote in a previous blog, this creates what I call the iPhone conundrum: people compare enterprise AI capabilities to the polished, seamless AI they experience as consumers, like on their smartphones, where things “just work.” This can be a challenge, as internal stakeholders may have unrealistic expectations or misunderstand the trade-offs involved in building AI for complex business processes. But you can also use this to your advantage: by tapping into familiar, intuitive examples from everyday AI, you can help make abstract concepts more tangible and inspire confidence that well-designed AI solutions can deliver meaningful value — even if they look and feel different inside the enterprise.

A key recommendation: avoid jumping into big group conversations too early. Focus first on engaging individual stakeholders, addressing their concerns, and building support one-on-one or in smaller settings. Large group meetings often risk derailing decisions, as individual concerns can quickly snowball into broader team doubts. By preparing and aligning key players ahead of time, you set the stage for more focused, productive group discussions when it really counts.

In the end, AI is just another feature and capability we as product managers can use to solve real business needs. We have a wide range of options at our disposal — from building custom analytical models, to leveraging proven industry models, to customizing existing tools on the market, or even simply turning on AI features already embedded within our business applications. As our business partners become more comfortable, confident, and educated about these capabilities, the conversations will get easier. Together, we’ll be able to focus less on the buzzwords and more on where and how AI can deliver meaningful, measurable value.