Bret Taylor on the AI Bubble and Sierra's Vision
This article summarizes a conversation with Bret Taylor, CEO of AI startup Sierra and chairman of OpenAI. He draws a compelling comparison between the present AI boom and the dot-com bubble of the late 1990s, emphasizing that both periods feature immense potential alongside the risk of significant financial losses. Taylor highlights Sierra’s unique business model, where clients pay only for successfully resolved customer service issues by AI agents, advocating for a shift towards outcome-based pricing in the software industry. He also offers insights into his evolving definition of Artificial General Intelligence (AGI), suggesting that current AI capabilities have already surpassed what was once considered AGI. The discussion delves into the transformative power of AI in various sectors, particularly customer service, and the challenges and opportunities that arise as AI models become more sophisticated and integrated into daily operations.
Detailed Report on the AI Industry and Sierra's Innovations
On September 11, 2025, during a live event in San Francisco, California, hosted by Alix Partners, Alex Heath, guest host of \"Decoder,\" interviewed Bret Taylor, the CEO of Sierra and chairman of OpenAI. Taylor, a seasoned tech veteran with a rich history at Google, Facebook, and Salesforce, shared his comprehensive vision for the future of artificial intelligence. His departure from Salesforce in early 2023 coincided with the public release of ChatGPT, which profoundly influenced his decision to venture into the AI space. Collaborating with Clay Bavor, Taylor co-founded Sierra, a company focused on developing AI agents to revolutionize business-to-consumer interactions, particularly in customer support.
Sierra’s distinctive revenue model charges clients solely when its AI agents autonomously resolve customer issues, such as adjudicating warranty claims or facilitating mortgage refinances. This outcome-based approach, Taylor argues, fosters a stronger alignment between software vendors and their customers, a departure from traditional licensing or hourly billing. He emphasized that this model is particularly beneficial for large enterprises, enabling them to handle a significantly higher volume of customer interactions at a fraction of the traditional cost. For instance, companies like ADT, Ramp, and SiriusXM are already leveraging Sierra's AI agents to manage customer calls and digital interactions, from routine inquiries to complex service upgrades.
Taylor revealed that voice-based AI interactions are becoming more prevalent on Sierra's platform than text-based ones, attributing this trend to the natural human inclination towards spoken communication and AI's ability to digitize traditional phone channels. He foresees a future where AI agents seamlessly handle tasks across various communication modes, making technology more accessible and integrated into daily life. From personal assistants managing emails during commutes to sophisticated agents navigating regulated industries like health insurance and banking, Taylor envisions a world where AI streamlines complex processes, improving efficiency and customer satisfaction.
Regarding the technical aspects of Sierra’s operations, Taylor clarified that the company does not pre-train its own large language models but instead focuses on fine-tuning existing ones. He stressed that applied AI companies should leverage these rapidly advancing models as infrastructure, selecting the most suitable ones based on speed, cost, and quality requirements for specific tasks. This approach allows Sierra to concentrate on developing robust applications with advanced features like guardrails for regulated processes, multilingual support, and intelligent noise cancellation, addressing complex challenges that current off-the-shelf AI tools cannot.
The conversation also touched upon the concept of Artificial General Intelligence (AGI) and superintelligence. Taylor noted that the definition of AGI has continuously evolved, with current AI capabilities already surpassing what was once considered AGI years ago. He now defines AGI as models exceeding human intelligence in almost all digital domains and ideas, acknowledging the distinction between intelligence and physical interaction. Taylor expressed optimism about AI's potential for scientific breakthroughs, citing recent advancements in AI-generated mathematics as an example. He concurred with Sam Altman's assessment of an AI bubble, likening it to the internet bubble where significant value creation coexists with financial speculation and failures. However, he distinguished Sierra's approach from projects that merely experiment with AI, emphasizing that Sierra provides concrete, outcome-driven solutions that deliver measurable business value.
The Transformative Power of AI: A New Era of Efficiency and Accessibility
Bret Taylor’s insights offer a profound look into the ongoing AI revolution, highlighting its potential to redefine how businesses operate and interact with their customers. His vision for AI agents, particularly through Sierra’s outcome-based pricing, signifies a pivotal shift towards accountability and tangible value in the software industry. This model not only promises economic benefits by reducing operational costs but also paves the way for enhanced customer experiences, making services more accessible and personalized. The growing importance of voice interaction with AI, as noted by Taylor, points to a future where human-computer interfaces become increasingly intuitive and ubiquitous, bridging digital divides and integrating seamlessly into our daily routines. While the parallels to past tech booms suggest a period of both immense opportunity and significant risk, Taylor’s focus on practical, problem-solving applications of AI offers a hopeful outlook. This approach underscores that true innovation lies not just in technological advancement but in its effective and ethical deployment to solve real-world challenges and benefit humanity.