Mar 03, 2026 · Product Momentum Podcast · 6 min read

AI Product Strategy: The Sense Shape Steer Framework

Bancy Maida introduces the Sense Shape Steer framework for AI product development. Learn how to navigate the blurring of UX and product roles, avoid the 'average mean' trap of AI-generated content, and define critical success criteria for high-stakes AI implementations.

Technology Business Strategy Product Management

The takeaways

Prioritize Problem Solving Over Role Boundaries. Shift team focus from individual role definitions to collective problem-solving. Start every AI initiative by asking what problem is being solved and why, ensuring alignment before technical execution.
Avoid the AI Average Mean Trap. Recognize that AI-generated prototypes often drive toward the statistical mean, creating mediocre solutions. Designers must actively raise the bar to push beyond average outputs and deliver genuine user delight.
Define AI Experience Modes Early. Explicitly categorize AI interactions as ambient, generative, or agentic during the 'Sense' phase. This clarity prevents scope creep and ensures the technical architecture matches the intended user experience.
Establish Contextual Accuracy Thresholds. Set specific accuracy targets based on risk and reversibility. High-stakes actions like medical prescriptions require 99% accuracy, while lower-risk tasks may tolerate 80-90% with human oversight.
Implement Rigorous AI Evaluation Metrics. Define AI evals and trust metrics before launch to prevent experiments from stalling. Use subject matter experts to pressure-test complex agentic workflows beyond standard QA capabilities.
Use Storyboards Instead of Figma First. Delay interface design in favor of user-perspective storyboards. Map user goals, context, and AI intervention points frame-by-frame to ensure the AI role is strategically integrated before visual design begins.

The Shift from Role to Outcome

The rapid advancement of AI tools has fundamentally altered product development workflows, blurring the traditional boundaries between UX designers, researchers, and product managers. As noted by Bancy Maida, CEO of Coro UX Design, the industry is moving away from role-specific silos toward a collaborative model focused on "what needs to be done." This shift is driven by the speed at which AI can generate high-fidelity prototypes, reducing the time from idea to code to mere days. However, this acceleration has created a critical vulnerability: the erosion of discovery patience. Teams are increasingly skipping deep research in favor of rapid implementation, leading to products that are technically functional but strategically misaligned with user needs.

The Average Mean Trap

A central challenge in AI-driven product design is the tendency of generative models to produce "average" outputs. AI is trained to predict the most likely next step, resulting in solutions that are safe, predictable, and often mediocre. This creates a cognitive trap for designers and stakeholders who may accept these prototypes as sufficient, thereby stifling innovation. Maida argues that the role of the designer is no longer about pixel-perfect execution or speed, but about "raising the bar" to push beyond the mean. The goal is to identify opportunities for genuine user delight that AI cannot autonomously discover, ensuring that products stand out in a market saturated with similar, AI-generated solutions.

The Sense Shape Steer Framework

To address these challenges, Maida introduces the Sense Shape Steer framework, a structured approach to AI product development. The Sense phase involves identifying the intersection of user needs and AI capabilities, defining the specific "experience mode" (e.g., agentic, generative), and establishing risk guardrails. The Shape phase moves away from immediate interface design in tools like Figma, instead utilizing storyboards to map user journeys and AI intervention points. Finally, the Steer phase focuses on implementation and evaluation, defining rigorous AI metrics and trust indicators before launch. This framework is particularly critical in high-stakes industries like healthcare, where accuracy thresholds must be contextualized based on the severity of potential errors. By defining success criteria early, teams can avoid the common pitfall of AI experiments that fail to transition to production due to a lack of thoroughness and clear evaluation metrics.

Strategic Implications

For product leaders, the key takeaway is that speed without structure leads to diminishing quality. Organizations must invest in frameworks that enforce critical thinking about AI's role, accuracy requirements, and user trust. The future of product management lies in hybrid skills where teams can prototype rapidly but also critically evaluate the strategic value and quality of AI-generated outputs.

Key insights

AI tools drive outputs toward the statistical mean, creating a risk of mediocre product quality if designers do not actively push for higher standards. The ease of prototyping has reduced industry patience for deep user research, leading to a gap between technical capability and user value.

Product Quality →

Impact: Teams that fail to raise the bar on AI-generated outputs will produce undifferentiated products that struggle to gain user adoption or loyalty.
The traditional role boundaries between UX, research, and product management are dissolving. Success now depends on cross-functional collaboration focused on problem-solving rather than individual task execution.

Team Dynamics →

Impact: Organizations that enforce rigid role silos will move slower and less effectively than those that adopt a fluid, outcome-oriented team structure.
AI implementation requires defining specific "experience modes" such as ambient, generative, or agentic. This classification is crucial for determining the appropriate technical architecture and user interaction patterns.

AI Strategy →

Impact: Failing to define the AI experience mode early leads to architectural mismatches and user confusion, increasing development costs and reducing product usability.
Accuracy thresholds for AI must be contextualized based on risk and reversibility. High-stakes actions require near-perfect accuracy, while lower-risk tasks can tolerate higher error rates with human oversight.

Risk Management →

Impact: Applying a one-size-fits-all accuracy standard leads to either unnecessary friction in low-risk areas or dangerous liability in high-risk areas.
AI experiments often fail to reach production due to a lack of defined evaluation metrics. Establishing AI evals and trust metrics before launch is essential for validating product viability.

Implementation →

Impact: Without pre-defined success criteria, AI features remain perpetual pilots, failing to deliver measurable business value or user trust.

Action items

Adopt the Sense Shape Steer framework for all new AI initiatives. Begin with the Sense phase to identify the intersection of user needs and AI capabilities, explicitly defining the AI experience mode and risk guardrails.

Impact: This structured approach ensures that AI features are strategically aligned with user problems before technical development begins, reducing waste and misalignment.
Replace initial Figma mockups with user-perspective storyboards. Map out user goals, context, and specific AI intervention points frame-by-frame to validate the AI's role in the journey before designing interfaces.

Impact: This prevents interface-first thinking from constraining AI capabilities and ensures that the AI's role is logically integrated into the user workflow.
Define contextual accuracy thresholds for each AI feature. Determine the required accuracy level based on the severity of potential errors and the reversibility of the action, documenting these as success criteria.

Impact: Clear accuracy targets allow engineering and QA teams to build appropriate fallback mechanisms and human-in-the-loop controls, enhancing user trust and safety.
Establish AI evaluation metrics and trust indicators prior to launch. Identify specific metrics for accuracy, adoption, and user trust, and engage subject matter experts to pressure-test complex agentic workflows.

Impact: Pre-defined metrics enable objective validation of AI performance, facilitating a smoother transition from experiment to production and ensuring the feature delivers measurable value.
Train design and product teams to critically evaluate AI-generated prototypes. Encourage teams to look beyond the "average" output and actively seek opportunities for user delight and differentiation.

Impact: This cultural shift ensures that AI is used as a tool for augmentation rather than a replacement for creative thinking, leading to higher-quality, more innovative products.

Quotes

“The conversation often needs to start from what are we solving and why?”

“It is designed to be that average mean, which means you will never look at it and think that, oh my god, this is so bad, or no, this doesn't work.”

“Quality is not volume.”