In the last 60 days, we've witnessed a remarkable shift. OpenAI's Sora 2 hit 1 million downloads in five days with a complete video creation platform including social features and identity insertion. Anthropic launched Claude Skills and industry-specific workflows for financial services and life sciences. Google expanded NotebookLM with video overviews, custom personas, and Mind Maps. These aren't model improvements - they're complete products putting LLMs in useful contexts. The product era of AI has arrived, and research from Procter & Gamble reveals exactly why this matters.
Harvard researchers studied 776 P&G professionals and discovered that individuals using AI performed as well as two-person teams without AI, and teams with AI were three times more likely to generate top 10% solutions. But this only worked when organizations stopped treating AI as a productivity tool and started treating it as a teammate. "Companies that focus solely on efficiency gains from AI will not only find workers unwilling to share their AI discoveries for fear of making themselves redundant but will also miss the opportunity to think bigger about the future of work," notes Wharton's Ethan Mollick, who co-authored the study. The AI-enabled teams also reported more positive emotional experiences, breaking down functional silos between technical and commercial professionals. What was the difference? Product thinking, not feature thinking.
Here's is a five-step process for reimagining user experiences with AI:
First, start with the user problem, not the technology. Sora 2 didn't begin with "better video generation" - it started with "how do people want to create and share video stories?" The result: Cameos that let you insert yourself into videos, remix features for collaboration, and a social feed for discovery. Claude Skills didn't ask "how can we improve Claude?" - it asked "how do financial analysts actually work?" and built Excel integration, real-time market data connectors, and pre-built DCF models.
Second, design AI as a collaborative teammate within complete workflows. NotebookLM doesn't just answer questions - it produces Audio Overviews, Video Overviews, Mind Maps, and Reports. It's grounded in your sources, customizable with personas, and integrated with Google Workspace. The product is the research workflow, not the chat interface.
Third, package domain expertise into context-specific applications. Claude for Financial Services scores 55.3% on finance benchmarks because it comes with industry-specific Skills - not because it's a better general model. Claude for Life Sciences integrates directly with Benchling and PubMed because that's where scientists actually work. Load the expertise when needed, stay efficient when it is not needed.
Fourth, build for compounding value and specific outputs. The P&G study showed AI doesn't just boost individual productivity - it creates knowledge sharing across teams and generates exceptional solutions. Sora 2's remix features mean every creation becomes a starting point for others. NotebookLM's multiple output formats (audio, video, mind maps) mean one research session serves multiple use cases.
Fifth, measure both performance gains and emotional experience. The P&G researchers found that AI-enabled workers were not just more productive - they were also happier, more engaged, and felt more capable across functional boundaries. Teams using AI reported positive emotions matching or exceeding traditional two-person teams. If your AI strategy makes work feel robotic or threatening, you're doing it wrong.
Ready to apply this to your business? Start by auditing your current AI initiatives against these five steps. Are you designing AI as a teammate or treating it as automation? Are you building complete workflows or just adding chat features? Are you packaging domain expertise or relying on general capabilities?
The product era of AI rewards those who think like product builders, not technology adopters.