With the advent of AI, what kind of apps will be needed??

When AI makes software cheap to build, competitive advantage shifts from the app itself to what can't be replicated: proprietary data, real-world trust, earned community, and proven outcomes. The question becomes not "what's addictive?" but "what has a defensible relationship to reality?" **Key points** • Apps with proprietary access (real-time data, regulated transactions, trusted relationships, built communities) survive commoditization. • Curation and taste layers become scarce — filtering infinite AI-generated options by what's actually good *for you*. • Outcome-oriented agents win by delivering results fast, not by maximizing time-on-app; engagement metrics invert. • ThinkThru's moat is the profile as intellectual fingerprint — a GitHub-like record of how someone actually thinks (steelmanning, shifting positions, anchoring in reality), earned through contributions, not self-reported. • The profile must feel like self-discovery, not performance review; it becomes the lock-in because it's irreplaceable proof-of-intellectual-work. **Where it landed** Pricing should tie to profile depth and contribution volume, not seats. Profiles should be public by default, capturing behavioral patterns across multiple threads before generating precise (not generic) characterizations. The test: would losing this account feel like genuine loss? **Still open** • How many threads must accumulate before an AI summary becomes meaningful to show a user? • Which specific numbers (pricing tiers, tag limits per session, minimum thread thresholds) actually represent group consensus versus plausible placeholders?