Brett Taylor

Co-founder of Sierra, Chairman of the Board at OpenAI, and former co-CEO of Salesforce.

112

Mentions

Appeared in

Discussed in

Key positions and views

AI agents are a new class of software that autonomously perform jobs, fundamentally changing the total addressable market for software from tool budgets to labor budgets.

The most effective and logical business model for AI is outcomes-based pricing, where customers pay for successful resolutions, directly tying the software's cost to its ROI.

Building proprietary foundation models is a strategic error for most applied AI companies due to extreme capital expenditure and the rapid commoditization of the underlying technology.

The future of creating software is not writing code but operating AI systems that generate code, which will require a new 'programming system' to manage.

Large, incumbent companies are disadvantaged in the AI transition not by a lack of technology, but by the difficulty of adapting their existing business models and go-to-market strategies to the new paradigm.

Podcast consensus on Taylor

Points of consensus

▶AI agents represent a fundamental shift in software, moving from productivity tools to autonomous systems that perform entire jobs, thereby expanding the total addressable market from software budgets to labor budgets.Apr 2026

▶Applied AI companies should not build their own foundation models. This is considered a poor use of capital due to the extreme expense and the rapid commoditization of the underlying model market, which will consolidate to a few large players.Feb–Apr 2026

▶The future of software development will be transformed from manual coding to operating AI-powered, code-generating systems, necessitating a new kind of 'programming system' beyond just a new language.Apr 2026

▶Outcomes-based pricing is the superior and inevitable business model for AI agents, as their value can be directly attributed to business ROI, such as resolving a customer service case.Feb 2026

Points of debate

▶Pace of Automation vs. Job Creation: Taylor consistently highlights AI's capacity to automate entire jobs in sectors like software engineering and customer service, leading to massive productivity gains. However, the claims do not address the countervailing force of job creation or the net societal impact of this rapid automation.Feb 2026

▶Commoditization of Models vs. Sustainable Moats: Taylor argues foundation models are 'fastest-depreciating assets' that are commoditizing, yet his company's success relies on them. A key tension is whether the 'tax' paid to model providers will erode margins for applied AI companies or if durable moats can be built through vertical expertise, data, and distribution.

▶Best-of-Breed Startups vs. Incumbent Power: Taylor posits that tech disruptions favor 'best of breed' solutions because incumbents are constrained by their business models. Simultaneously, he warns that large infrastructure providers like AWS and Azure pose a significant threat to AI tooling companies, creating a debate about whether startup agility or incumbent scale will ultimately dominate the application layer.Feb 2026

▶AI Safety Through Iteration vs. Existential Risk: Taylor advocates for responsible, iterative deployment as the primary mechanism for ensuring the safety of future, more powerful AI models like a hypothetical GPT-8. This practical, market-driven approach contrasts with more cautious perspectives that emphasize the need for proactive, theoretical alignment research to mitigate potential long-term risks before deployment.

Key themes

▶The Agent as a New Software ParadigmFeb–Apr 2026

Taylor consistently argues that AI agents are not just an evolution of software but a new category altogether. He defines them as autonomous systems capable of performing entire jobs, contrasting them with traditional software that merely enhances human productivity. This shift, he believes, makes a company's branded AI agent as critical as its website or mobile app.

For investors, this theme reframes the total addressable market for enterprise software, shifting the valuation calculus from per-seat licenses tied to employee headcount to outcomes-based models tied to entire labor market budgets.

▶The Economics and Business Models of AI

A core theme is the unique economic structure of the AI industry. Taylor predicts the entire software market will adopt outcomes-based pricing, where customers pay for results (e.g., a resolved support ticket) rather than usage. He also emphasizes the bifurcation of the market, with a capital-intensive, consolidating foundation model layer and a more capital-efficient applied AI layer.

This suggests that long-term value capture in the AI stack will be less about owning the 'best' model and more about mastering a specific business vertical and proving direct, quantifiable ROI to customers.

▶The Transformation of Software CreationFeb–Apr 2026

Taylor foresees a radical change in how software is built, moving from developers manually writing code in IDEs to operating AI systems that generate the code. He believes this shift is so profound that it requires a new 'programming system,' not just a better language, and identifies professions like software engineering as being among the most impacted by AI.

This signals a potential shift in the value of technical skills, where expertise in system design, AI orchestration, and context management may become more critical than proficiency in a specific programming language.

▶Market Disruption and Incumbent DisadvantageFeb 2026

Drawing on historical examples like SGI and Sun Microsystems, Taylor argues that major platform shifts, like the current AI wave, are difficult for large incumbent companies to navigate. He asserts their primary obstacle is not a lack of technological capability but the rigidity of their existing business models and go-to-market strategies, which creates opportunities for 'best of breed' startups.

Analysts should scrutinize incumbents' AI strategies not just for technical innovation but for fundamental changes to their pricing, sales motions, and service delivery models, as these are the likely points of failure or success.

Source episodes

Sentiment over time

Not enough data for timeline

Changes over time

Early 2000s

As a product manager at Google, Taylor led the V2 of Google Local, which evolved into the highly successful Google Maps. This period was marked by rapid product innovation, such as the integration of satellite imagery that attracted 90 million users on its first day.

Late 2000s

Co-founded the social aggregation service FriendFeed. This experience provided a direct lesson in market dynamics and distribution, as FriendFeed lost momentum to Twitter after the latter successfully recruited high-profile celebrities to its platform.

Early 2010s

Served as the Chief Technology Officer at Facebook. During his tenure, he grappled with major technological challenges, including a poorly rated iOS app, and learned the importance of executive dogfooding to improve product quality.

Mid-2010s to Early 2020s

Founded the productivity company Quip, which was acquired by Salesforce for $750 million. This led to his ascent within Salesforce, culminating in the role of co-CEO, where he gained deep experience in enterprise go-to-market strategies.

November 2023

Stepped in as Chairman of the Board at OpenAI during its leadership crisis. His stated goal was to stabilize the organization, which involved reinstating Sam Altman as CEO and initiating an independent review of the preceding events.

2022 - Present

Co-founded Sierra, an AI agent company for enterprises. Under his leadership, the company has experienced hyper-growth, reportedly reaching $150 million in ARR in eight quarters and a $10 billion valuation, championing an outcomes-based pricing model.

Suggested prompts

How does Bret Taylor's 'outcomes-based pricing' model challenge traditional SaaS valuation metrics, and what are its potential long-term risks for revenue predictability? &nearr;Taylor argues incumbents fail due to business model constraints, not technology. Analyze the AI strategies of Salesforce and Microsoft through this lens; do they confirm or contradict his thesis? &nearr;Evaluate Taylor's prediction of a 'constellation of models.' What are the technical and business implications for applied AI companies like Sierra if this heterogeneous, multi-model future materializes? &nearr;Taylor's career spans web, social, mobile, and AI. How have his experiences at Google, Facebook, and Salesforce informed Sierra's strategy, particularly its focus on enterprise go-to-market over product-led growth? &nearr;

Key concepts

AI Agents 6 ep Sierra 6 ep Outcomes-Based Pricing 5 ep Foundation Models 4 ep Go-to-Market Strategy 4 ep Customer Service Automation 4 ep Future of Software Development 3 ep Market Disruption & Incumbents 2 ep OpenAI 2 ep Google Maps 1 ep

Notable quotes

“these AI agents actually are doing jobs rather than just adding productivity enhancements on someone else doing the job. And I think that value is so quantifiable and so real that people will pay a lot more for it than you will a productivity enhancement.”

Bret Taylor · Bret Taylor’s Journey Leading Salesforce, Sierra & OpenAI

“it turns out that spending tens of millions of dollars to train your own model when the foundation model market is sort of commoditizing, turns out to be a really unintelligent use of capital, at least in the applied AI market.”

Bret Taylor · Bret Taylor’s Journey Leading Salesforce, Sierra & OpenAI

“I think we'll probably see the first trillion dollar software as a service company come out of this era. And a pure play, not building infrastructure and things like that.”

Brett Taylor · Episode 250: AI Special Featuring Sierra, Harvey, Windsurf & More

“the main reason GPT-8 will be safe is because GPT-1 through 7 were safe through responsible iterative deployment, and we as a society grapple with the impact.”

Bret Taylor · Bret Taylor’s Journey Leading Salesforce, Sierra & OpenAI

Report last updated: Apr 21, 2026

Get started free

Back to Entities Intelligence Report