Taylor draws parallels between the AI revolution and the early internet, noting that both feature obvious, high-value markets (like search then, customer service now) leading to intense competition where execution and business model, not just technology, determine the winner.
Incumbent SaaS giants face a significant challenge in adopting AI, not because of technology, but because of their established business models, sales incentives, and investor expectations. Taylor argues that the shift to outcome-based pricing is a formidable change-management problem that will create opportunities for new, unencumbered companies.
AI agents represent a fundamental shift from software as a productivity tool to software that performs a job. Consequently, their value should be measured based on the outcomes they produce (e.g., a sale made, a problem solved), not on a per-seat license, which will upend traditional SaaS metrics and business models.
Taylor predicts the AI software stack will consist of foundation model providers selling intelligence as infrastructure, and applied AI companies building purpose-built agents for specific industries and functions. He believes most enterprises will prefer to buy these specialized agents rather than build them, due to the ongoing costs of maintenance and the value of amortized innovation.
Keep pulling the thread on Bret Taylor.