Your Own Customers Will Disrupt You With AI

Key Takeaways

• Strava’s AI gave generic praise despite data showing slower performance
• Claude extracted actionable insights from raw race and heart‑rate files
• Users can build custom AI tools using platform data in days
• Incumbents must personalize AI using existing user history to stay competitive
• Agile labs groups are essential for rapid AI product innovation

Pulse Analysis

The rise of generative AI has turned data ownership into a competitive weapon. Platforms such as Strava and Duolingo have accumulated years of granular activity logs, heart‑rate curves, and language‑learning patterns, yet many still surface generic, one‑size‑fit messages. When a user feeds the same raw datasets into a large language model like Claude, the model can produce detailed performance diagnostics, tailored workout suggestions, or personalized language tutoring within minutes. This contrast highlights a widening gap: the technology to interpret personal data exists, but the product layer often fails to activate it.

That gap creates a paradoxical threat: the most valuable customers become the most capable competitors. In the article, the author shows how a weekend‑level prompt to Claude revealed neuromuscular fatigue that Strava’s AI ignored, while a former Duolingo power user built a Spanish tutor that outperformed the commercial app. Because the data is already locked into the incumbent’s ecosystem, a motivated user can extract it, apply an off‑the‑shelf model, and ship a superior experience for a fraction of the cost. The barrier to entry is now knowledge, not capital.

Incumbents must flip the script by turning their data advantage into hyper‑personalized AI services. Building small, autonomous labs with direct profit‑center mandates allows rapid experimentation without legacy product constraints. These teams should integrate user‑specific signals—training history, injury notes, language errors—into model prompts that generate context‑aware feedback in real time. By delivering insights that feel uniquely “seen,” companies can reinforce lock‑in, justify subscription fees, and neutralize the DIY disruption pipeline. Failure to adopt this user‑centric AI approach risks losing both relevance and revenue to the very customers they serve.