Why the Next Hit AI Product Will Be Social
the gist
Sarah Tavel argues that current AI products are trapped in a 'single-player' paradigm and that the next wave of consumer AI winners will be built by product-minded founders who integrate social dynamics, status, and network effects to help users learn from one another.
The Shift from Technical to Product-Centric Founders
Sarah Tavel observes a recurring cycle in technology: early-stage paradigm shifts are dominated by deeply technical founders who build the necessary infrastructure, while later-stage maturity favors 'product geniuses' who focus on user experience and engagement. She likens the current state of ChatGPT to Google’s early days—a powerful, backend-heavy tool that is essentially a text box. While OpenAI’s research-first approach was the correct bet for this specific paradigm shift, Tavel believes the industry is now moving toward a phase where the underlying models are stable enough that the competitive advantage will shift to those who can design superior, more intuitive user interfaces.
The Missing Social DNA in AI
Despite the success of LLMs, Tavel finds current AI products to be largely 'single-player' experiences. She argues that the lack of social infrastructure is a missed opportunity. Users are currently forced to reinvent the wheel, manually crafting prompts and custom instructions to get the results they want. Tavel envisions a future where AI products incorporate 'social DNA'—features that allow users to follow experts, share prompts, and build status within a community. This would transform AI from a solitary tool into a collaborative network where the collective knowledge of power users makes the technology accessible to the average person.
Designing for Trust and Community
For a social AI product to succeed, it must solve the problem of trust and discovery. Currently, platforms like ChatGPT’s 'GPT Store' lack transparency; users cannot see the underlying prompts or documents that make a specific GPT effective. Tavel suggests that the next generation of AI products will need to provide visibility into how these tools are built, allowing users to vet the 'AI personalities' they adopt. She draws a parallel to the 'you are the average of the five people you spend time with' concept, suggesting that users will soon curate a set of AI personalities—some for work, some for personal wellness—that act as their own personal board of advisors.
Defining Real Network Effects
When evaluating startups, Tavel warns against 'flywheel diagrams' that lack substance. She defines real network effects in this context as the presence of status-seeking behavior. In a successful social AI product, participants should be incentivized to contribute high-quality prompts or configurations because doing so earns them status, celebrity, or authority within the network. Without this incentive structure, a product is merely a utility rather than a platform with true network effects.