There’s a myth in modern business that innovation has become self-driving—that once you feed enough data into AI, it can do the rest. No customers, no designers, no mess—just cold, beautiful efficiency.

But the companies actually leading the charge don’t see it that way.

In our survey and interviews with firms listed in Fast Company’s 50 Most Innovative Companies, 89% of companies responding said they “often” or “always” prioritize understanding and addressing customer needs in their product development process. And 72% said they actively integrate direct user feedback and empathy-driven insights throughout the journey. 

The message from this survey and our interviews with executives at those companies is clear: In an age of generative everything, the most innovative companies are still grounded in human understanding. Sophisticated innovators aren’t replacing empathy with data—they’re engineering ways to scale it.

Creating and launching new ideas is easier than ever today, but with that abundance comes a new challenge: ensuring that what you build truly matters to real people. The key is blending the speed and scale of AI with the depth and nuances of real customer feedback. The strongest innovators use AI to enhance—not replace—their direct engagement with users. Innovation leaders are combining AI with real-world input to validate faster, iterate smarter, and build solutions people genuinely want.

Synthetic personas come with rewards ... and risks

Nowhere is this more evident—or more debated—than in the rise of synthetic personas: AI-generated user archetypes trained on massive data sets, capable of simulating human behaviors, choices, even frustrations.

Used wisely, synthetic personas can unlock enormous value by:

• simulating hard-to-reach or emerging user segments before they exist in the market;
• rapidly pressure-testing ideas across dozens of theoretical contexts;
• capturing patterns, behaviors, and emerging needs across vast audiences with remarkable precision;
• simulating real user experiences at scale—enabling faster, more informed innovation decisions; and
• reducing time-to-insight in early-stage product design or personalization efforts.

Synthetic personas are powerful, but they're not substitutes. They work best alongside real users, not instead of them:

• They reflect probabilistic behavior and thus may miss nuance, emotion, or edge cases.
• Overreliance can create false confidence in products that haven’t been validated with real people.
• They can reflect bias embedded in their training data, scaling inaccuracies faster than traditional personas.

Bottom line: Synthetic personas are force multipliers, not foundations. They should supplement, not supplant, direct engagement with real customers.

People still provide the strongest innovation signals

When we asked the Fast Company innovators what tools or techniques they use to identify trends or crowdsource ideas, more than two-thirds said they relied on analysis of customer needs and competitive analysis, while well over half said they were monitoring regulatory and societal trends.

By contrast, significantly less than half reported using internal crowdsourcing, relying on market analytics or external open innovation, or using venture funding technology analysis.

Despite the hype, the loudest signals still come from listening—not modeling.

Leaders invest in AI that accelerates empathy

When asked where they plan to invest over the next five years, the Fast Company innovators we interviewed pointed overwhelmingly to AI-powered tools that increase velocity without compromising human focus:

• AI-assisted customer research and trend scouting
• Persona simulations, ideation, and concept generation
• Rapid pretotyping, prototyping, and A/B testing
• AI-driven design, no/low-code, and proof-of-concept acceleration
• Brand and marketing messaging for rapid experimentation at scale
• Automation of low-value work to free up time for customer engagement

Recent work with one of our own clients shows how companies can move from exploring product opportunities to validated concepts in a matter of hours by using a combination of AI-based tools and real consumer inputs.

In this case, the company identified opportunities by using AI-based social listening tools that could gather consumer insights at speed. It then used a custom GPT to generate ideas and mocked up concepts with GPT and AI image generation tool Midjourney. (In the case of digital products, companies can even create full-scale prototypes using UX/UI automation tools.)

The company then tested the AI-generated concepts with real consumers across key target segments, using Outset.AI to moderate interviews and synthesize and prioritize concepts. The result: fast qualitative research results across a high sampling of participants in just a matter of hours. These results would have taken weeks via traditional methods, but AI tools were able to accelerate them without sacrificing closeness with real consumers.

This is the new philosophy of innovation technology: It should reduce the noise so that teams can hear the customer more clearly. AI isn’t replacing design thinking; it’s making it faster, deeper, and more scalable.