Eliza Wong
At the 2025 Bund Summit, Zhu Xiaohu (朱啸虎), the renowned managing partner of GSR Ventures (金沙江创投), cut through the AI hype with a stark reminder: flashy technology doesn’t pay the bills—user retention does. In a landscape saturated with futuristic promises, his message was clear: the most investable AI companies solve real problems for real users, day after day. For investors and entrepreneurs alike, understanding this distinction isn’t just beneficial—it’s essential for survival in the rapidly evolving AI market. Zhu’s emphasis on user retention as the make-or-break metric offers a practical lens through which to separate fleeting trends from enduring businesses. Here’s a deeper look at his insights and what they mean for the future of AI investment.
– User retention is the most reliable indicator of an AI company’s long-term viability, surpassing technical novelty or hype.
– Practical, less ‘sexy’ AI applications—like meeting transcription tools—are currently outperforming more experimental technologies in commercialization.
– AI cannot fully replace humans in complex, high-stakes systems until the hallucination problem is resolved, creating opportunities for specialized solutions.
– The development of AI super-apps or dominant platforms is inevitable but will require deeper integration with specific user scenarios and needs.
– Real-world adoption and daily utility, not technological brilliance alone, determine which AI companies succeed and attract further investment.
Why User Retention Matters More Than Technological Brilliance
In the world of AI investing, it’s easy to be dazzled by cutting-edge demos or breakthrough algorithms. However, Zhu Xiaohu (朱啸虎) argues that sustainable business value isn’t built in research labs—it’s built through consistent user engagement. When judging AI companies, he insists that user retention is the single most important metric. Early adopters might flock to novel features, but lasting adoption depends on delivering reliable utility.
The Pitfalls of ‘Cool’ vs. ‘Useful’ AI
Many startups fall into the trap of prioritizing technological sophistication over practical applicability. Zhu notes that technologies which appear less revolutionary often have clearer paths to monetization. For instance, while generative AI models captivate imaginations, meeting transcription services like Abridge and Plaud.ai are already generating significant revenue. These tools leverage stable, well-understood AI capabilities to address everyday pain points—proof that commercial success doesn’t require solving artificial general intelligence.
Learning from 2024’s Top AI Commercialization Stories
Zhu pointed to meeting transcription apps as standout examples of AI done right. These platforms don’t rely on speculative tech; they use speech recognition and natural language processing to save professionals hours of manual work. Their value proposition is immediate and measurable, which translates into strong user retention.
Case Study: The Rise of Plaud.ai
Plaud.ai, now valued at over a billion dollars, exemplifies this principle. Unlike many AI startups burning cash on customer acquisition, Plaud.ai focused on refining a product people actually use repeatedly. Its growth wasn’t driven by viral marketing but by organic, recurring usage—a classic sign of product-market fit. Zhu highlighted that such companies often fly under the radar initially because their solutions seem obvious in hindsight. Yet, that very obviousness is what makes them scalable and sustainable.
Why AI Still Can’t Replace Humans—And What That Means for Investors
Acknowledging AI’s limitations is crucial for rational investing. Zhu emphasized that as long as AI systems produce errors or ‘hallucinations,’ they cannot be trusted with mission-critical processes. This isn’t a short-term hurdle; it’s a structural constraint that defines where AI can and cannot dominate.
The 1% Problem: Hallucinations and High-Stakes Scenarios
In fields like healthcare, finance, or complex software engineering, even a 1% error rate is unacceptable. AI might handle routine coding tasks, but designing intricate systems requires human oversight. This reality creates investable opportunities in ‘human-in-the-loop’ AI solutions—tools that augment rather than replace human expertise. For investors, backing companies that acknowledge this balance is smarter than betting on full automation.
The Future of AI: Super-Apps, Smart Agents, and Strategic Patience
Zhu believes super-apps will eventually emerge in the AI space, but their development will be gradual and context-dependent. He drew parallels to the mobile internet era, which produced only a handful of giant winners. Similarly, AI’s breakout applications will likely arise from deep integration with specific industries or user habits.
Why Siri Isn’t the Super-Entry Point—Yet
Despite Apple’s resources, Zhu expressed disappointment in Siri’s AI capabilities, noting that it falls short of being a truly intelligent assistant. This gap illustrates that technical prowess alone doesn’t guarantee market leadership. Future super-apps will need to excel in reliability, personalization, and utility—areas where user retention will again be the defining benchmark.
Putting User Retention First in Your AI Investment Strategy
For investors, Zhu’s advice is straightforward: look beyond the hype and focus on adoption metrics. High user retention indicates product-market fit, sustainable monetization, and potential for scalability. It also suggests that the company has prioritized solving real problems over chasing trends. In a field as dynamic as AI, this discipline is what separates fleeting successes from enduring enterprises.
As AI continues to evolve, the companies that thrive will be those that deliver tangible, repeatable value to their users. Zhu Xiaohu’s insights remind us that in technology, as in business, fundamentals still matter most. For investors evaluating AI opportunities, let user retention be your guide—it’s the metric that never lies.
