Working with artificial intelligence (AI) agents often reminds me of the Amelia Bedelia stories I read as a child. You provide AI with instructions, and it interprets them.
Amelia Bedelia is a housekeeper who takes the instructions she is assigned very literally. For example, when told to “put the lights out,” she took the light bulbs outside. And when asked to “draw the drapes,” she drew a picture of the drapes. The charm of the books is the reader’s understanding of what those tasks look like in execution, in contrast with Amelia’s misunderstandings. She’s missing the context — why she’s being asked to do a task — and confidently takes action without asking clarifying questions.
This can happen with AI, too.
AI understands that a task needs to be completed, but it lacks the context to determine whether the task it’s completing is the correct one. An agent doesn’t understand whether what it’s doing will be helpful to the user. And it’s missing a broader vision of how the completed task will impact users. It is up to the end user to know if AI output is helpful, based on what they were expecting and the end goal they are trying to achieve.
Thank you, Amelia Bedilia: wrangling AI agents
As generative models drive development, the tech landscape in 2026 is flooded with code. Anyone with a prompt can build a feature, but it takes a product manager to prevent AI from generating goofy clutter, breaking the user experience, or worse.
Product managers’ responsibilities extend beyond overseeing output, supporting developers, and delivering features. Increasingly, product manager roles include establishing clear guidelines and managing context for AI.
Demand for product managers in an AI era
Product management is essential in this era of AI development. While AI can generate a huge amount of code, product managers are responsible for auditing, reviewing, and improving results to address errors, misinterpretations, and (sometimes) messes. Where Amelia Bedelia tripped on terminology with comical results, missteps with real-world AI can carry serious consequences. Product managers play a vital role in shaping product vision, design, and strategy.
So, what do product managers do?
A product manager sits at the nexus of technology, business, and user experience. Their leadership is cross-functional, and involves both proactive monitoring and establishing guardrails to leverage technology to expand research capabilities and enhance efficiency. Here’s what to expect from a product manager in 2026:
- Product strategy and competitive analysis
- Mockups and iterative ideation
- High-impact production requirement documents
- Stakeholder and subject matter expert (SME) synthesis
Product strategy and competitive analysis
Product managers looking to make the most of AI in their roles should lean heavily on responsibility and strategic implementation.
Responsibility: Product managers take big-picture vision and turn it into actionable, market-informed roadmaps. Their role is critical for defining the value a product or release brings to a target audience, from competitive benchmarks to gap analysis and strategic roadmapping. Product managers establish both short- and long-term product goals aligned with business objectives.
Leveraging AI: AI is capable of conducting extensive research leveraging public documentation. It’s like deploying an enthusiastic analytical investigator with the ability to gobble up and synthesize thousands of disparate resources. Once that data is “swallowed,” you can query it, and leverage AI to understand the competitive landscape by:
- Turning siloed information and knowledge into actionable data
- Identifying where your product will differentiate from current competition
- Understanding how your product can fill a gap in the market
Mockups and iterative ideation
Defining success will remain the paramount responsibility of a product manager in 2026 and beyond.
Responsibility: It’s more important than ever to create clear and concise product requirements — as we manage both humans and agents. What the product should look like and how it works is still something that needs to be clearly defined for the human user experience.
Leveraging AI: AI is a strong tool to leverage for challenging assumptions and testing your product.
- Create testable product mockups to gather feedback. In design-driven development, leverage AI to create wireframe-based prototypes. Then, use those prototypes with end-users to gather insights, test usability, and test performance. Teams can then use feedback to iterate and ensure the final product meets customer needs and delivers market value.
- Create sample applications to find market fit. While enterprise-level software should always have a software engineering team behind it for architecture, security, and compliance, prototypes can mimic the experience without the full development budget. Creating a software prototype with faux data can be very agile in its purpose for identifying product specifications early.
By the numbers, AI is enhancing product team performance, therefore replicating some of the key benefits of human-centered teamwork. In a recent study with Procter & Gamble, researchers at Harvard Business School found that AI reduced the time it took for individual employees and teams at P&G to develop innovative ideas by more than 13%.
High-impact product requirement documents
Product managers own both the product’s purpose and scope, overseeing the space between development execution and business strategy.
Responsibility: Strong documentation is fundamental to building the right thing in the right way.
Leveraging AI: AI can provide a second pair of eyes for product requirement documents (PRDs), and can be helpful in identifying both helpful insights and blind spots.
- Use AI to help review PRDs. Generate a list of product questions the product manager can answer to fill in any unidentified gaps. Think of it as a pressure test. Answering questions can make written requirements clearer, ensuring the right product is built and the target audience’s needs are met.
- Treat AI as a peer reviewer. Provide AI with background on product development specifications. Prompt it with specific questions focused on validating your work, for example:
- What am I missing?
- What can I add to make this clearer?
- If you were a developer, what questions would you have?
- Ask AI to write a summary of PRDs to ensure they match product expectations. Utilizing AI to build stronger documentation speeds up development time. Consider adding context that might have been overlooked and highlighting important details so developers can stay in flow state and continue working at a high velocity while creating the end product.
Stakeholder and SME synthesis
When working with key stakeholders and subject matter experts (SMEs), product managers become code breakers, translating specialized knowledge into sound product strategy.
Responsibility: Product managers are responsible for closing knowledge gaps, turning vision into technical clarity, digging into the reasons behind requests, and making sure everyone is clear and aligned on priorities.
Leveraging AI: AI can be used to help synthesize SME expertise and stakeholders’ vision, helping strip away noise and creating a clear path forward.
- Generate customer questions. Use it to help come up with a wide range of questions, allowing the product manager to focus on identifying which questions will be most impactful.
- Search for insights. Feeding an agent the transcripts of customer interviews and conversations with stakeholders can allow you to derive valuable insights from a session just moments after it ends, saving you from manually parsing through hours of recordings and pages of notes.
- Summarize discussion findings. Leverage AI to quickly summarize meetings and develop a list of key takeaways and action items.
If Amelia Bedelia were an AI agent, then it would be up to the product manager to ensure that she had clear instructions to build and work off of. In the AI era, product managers will continue to own the human elements of product development when it comes to vision, design, and strategy. Some real magic can happen when we use AI as a partner in research, to gut check our bias, test our tools, and accelerate synthesis.