The speakers frame the current AI wave not as just 'better software,' but as a new technological layer compounding on previous shifts like the PC, internet, and mobile. This allows for unprecedented speed of adoption, as seen with ChatGPT, and creates opportunities for new 'best of breed' companies to emerge.
The core primitive of the AI era is the 'agent'—autonomous software that can reason and complete tasks. The speakers predict agents will become more important than websites or apps for customer interaction, fundamentally re-architecting the internet's economy around demand generation and fulfillment.
Sierra is pioneering an 'outcome-based' pricing model, charging customers only when an AI agent successfully resolves an issue. This moves beyond seat-based or consumption-based pricing to directly align the vendor's revenue with the value delivered, shifting risk from the customer to the vendor.
The AI market is stratifying into three layers: a small number of capital-intensive frontier model labs (e.g., OpenAI), a tooling layer (e.g., vector databases), and applied AI companies (e.g., Sierra). The speakers argue that most applied companies should not build their own foundation models due to the rapid depreciation of these assets.
Operating in the AI era requires a fundamental shift in how companies work. This includes aggressively adopting AI tools like Cursor for coding, focusing on providing AI with the right context rather than just fixing its output, and having difficult conversations about how every role can be augmented or automated.
Keep pulling the thread on Bret Taylor & Clay Bavor.