Vishal Kapoor’s 10 Rules for Building Honest Products with AI | Figma Blog (opens in new tab)

AI product development is ultimately a trust challenge, not merely a technical one. Vishal Kapoor argues that AI should accelerate exploration and execution without replacing human judgment, empathy, or accountability. His approach centers on building products that remain transparent, secure, emotionally aware, and honest—especially in sensitive areas such as personal finance.

Start with First-Principles Thinking

• Break complex problems into their fundamental components before reaching for an AI solution.
• AI can accelerate ideation and iteration, but it cannot replace human intuition, taste, or a distinctive product perspective.
• Question basic assumptions to uncover better alternatives. For example, Affirm challenges why customers receive three payment-plan options rather than one, five, or a customizable plan.
• Thoughtful disagreement among people remains essential for generating meaningful insights; AI is best used to explore possibilities more quickly.

Stay Close to Human Emotions

• Product teams should regularly observe customers, conduct UX research, read app-store reviews, monitor social media, and speak directly with users.
• Metrics and dashboards identify patterns, but they do not fully explain the emotions behind customer behavior.
• Financial products especially require sensitivity to anxiety, frustration, trust, and relief—not just transactional outcomes.
• Affirm uses an internal AI tool called Pluto to investigate recent customer disappointments, while still relying on human observation and empathy to interpret those experiences.

Treat AI as a Teammate

• AI is neither a guaranteed productivity multiplier nor an inevitable replacement for employees; it is another participant in a collaborative product-development process.
• Tools such as Figma Make help teams convert customer insights into prototypes and test ideas faster.
• AI can audit large numbers of screens and interaction patterns across web, mobile, and desktop experiences, identifying outdated or inconsistent designs.
• Moving repetitive auditing and prototyping work from engineers to designers and product managers increases iteration speed and creates more room for creativity.

Test the Edge Cases

• Trustworthy products cannot be designed only around the happy path.
• Teams should deliberately explore unusual inputs, failure modes, and unexpected customer situations rather than assuming normal usage.
• The article begins this rule by emphasizing that authentic product quality depends on examining the difficult and overlooked scenarios where users are most likely to encounter confusion or harm.

The overall recommendation is to use AI aggressively for exploration, prototyping, and repetitive analysis—but keep humans responsible for defining the problem, understanding customers, challenging assumptions, and ensuring the final product is honest.