App Layer Wins AI
AlbertB@albert
+3.4%
since Oct 6
The idea
I believe AI models are becoming commodity infrastructure, like a cloud bill, while the real profit sits with applications that own the customer relationship, distribution, and proprietary data. So this basket splits between large distribution incumbents bolting AI onto products people already pay for, and vertical specialists sitting on workflows and data a model can't easily copy, plus a smaller sleeve of AI-native apps turning the technology directly into revenue.
What it owns
- Established software companies can layer AI features on top of products customers already depend on, turning model costs into a margin opportunity rather than a threat to their core business. · MicrosoftIt owns the distribution channel through Office and enterprise relationships, allowing it to charge for AI features regardless of which model provider it uses underneath., IntuitIt holds years of tax and financial data that models cannot replicate, and can upgrade its models without customers noticing because they remain locked into its workflows., ServiceNowIt owns mission-critical workflows and data at large companies, so AI makes its platform more valuable to customers already dependent on it., AdobeIt integrates AI into creative tools where professionals already spend their time, potentially keeping customers paying for AI features rather than switching to standalone models., SalesforceIt controls customer records and the daily workflows of sales teams, positioning itself to deepen those relationships with AI agents rather than losing customers to startups., SAPIt operates as the system of record for how large enterprises run, holding data and processes that general models cannot simply walk in and replicate.32%
- Specialist software companies sit on industry-specific data and workflows that general-purpose models cannot easily replicate, creating defensible positions as AI features expand their products. · Veeva SystemsIt holds regulated pharmaceutical workflows and development data that foundation models lack, positioning itself to sell specialized AI features to drug makers who cannot replicate that knowledge., Tyler TechnologiesIt runs software local governments depend on with high switching costs, creating decades-long customer relationships that can be deepened as AI features arrive., Guidewire SoftwareIt operates core insurance systems that carriers cannot remove, with claims data that is proprietary fuel for AI features those insurers cannot get anywhere else., SamsaraIt collects physical-world data from fleets and job sites that foundation models do not have access to, creating a moat for AI features built specifically for those verticals., ToastIt functions as the operating system for restaurants, owning payment and order data, positioning itself to sell AI tools tailored to restaurant operations., DoximityIt owns the physician network and engagement data that pharma advertisers value, creating a distribution advantage that models cannot replicate., Procore TechnologiesIt holds construction workflows and project data too specialized and complex for general models to understand, creating a foundation for AI features in that vertical.27%
- Application companies are already turning AI into recurring revenue, demonstrating that value accumulates at the app layer where customers pay for results, not compute. · ShopifyShopify distributes AI tools to millions of merchants already running their business on its platform, making distribution the advantage while AI features drive deeper engagement rather than enabling merchants to leave., DuolingoIt turns cheaper models into paid subscription features, widening margins as model costs fall while subscribers pay specifically for AI-enhanced learning experiences., PalantirPalantir builds the application layer on top of whatever models exist, switching between models while keeping customer relationships intact, with growth depending on commercial AI deal velocity., AppLovinIt operates an ad targeting engine fueled by proprietary data, converting AI into immediate revenue rather than relying on future monetization.20%
- Apps with massive user bases benefit when cheaper models improve their products and lower costs, keeping users engaged while the economics of AI models shift in their favor. · Meta PlatformsIt controls apps billions of people use daily, so AI improvements to ads and products translate directly to better economics as model costs decline., SpotifyIt owns the listener relationship and embeds AI features like recommendations inside an app people already subscribe to, retaining customers as the product improves.12%
- Cash9%