The traditional SaaS model, where a product is built once and sold multiple times at a low marginal cost, no longer applies to AI native apps. Every interaction with an AI product consumes tokens, which use energy and incur costs. This creates a pricing dilemma, particularly when dealing with light and power users who burn tokens at different rates within the same pricing tiers.

A potential solution to this problem is the Tokens as a Service (TaaS) model. TaaS involves software that adds value to tokens and is resold directly to users as a service. Examples of TaaS include AI video creation, research tools, and chatbots, which all place a value layer over an API key and sell the tokens for significant markups.

The key to success in the TaaS model is to add as much value as possible to the AI tokens. This can be achieved by leaning heavily into software development (90%) and using AI tokens sparingly (10%). By offloading the costs of value delivery onto free executable code, companies can minimize their expenses and maximize their markups. The most effective applications will be those that use code to solve problems and AI to add a magical layer on top, rather than relying solely on AI to brute-force a solution.

Photo by by Natallia on Pexels
Photos provided by Pexels