Bret Taylor

It has been such a privilege to partner with @HelloMinted at Sierra - creating an AI agent as personalized as their service and helping millions express themselves. Seeing all the personalized cards on our mantle is my favorite part of the holiday season

Live events are an emotional business. Vivid Seats built an AI agent to match that energy at every step of the fan journey — and in four weeks went from kickoff to an experience their fans actually love sierra.ai/customers/vivi…

I am proud of Sierra's partnership with GoFundMe on their new Smart Fundraising Coach. Based on these early results, GoFundMe expects to see an additional $125 million in help raised this year gofundme.com/c/newsroom/gof…

Excited to announced Sierra's newest office in Madrid, Spain 🇪🇸 sierra.ai/blog/building-…

One of Tony Hoare's quotes is particularly relevant in the age of agentic coding: "There are two ways of constructing a software design: one way is to make it so simple that there are obviously no deficiencies and the other is to make it so complicated that there are no obvious deficiencies."

Tony Hoare, the Turing Aware winner who (among many other things) invented quicksort and communicating sequential processes (CSP), which inspired the Go programming language, has passed away. RIP to one of the greats of our industry blog.computationalcomplexity.org/2026/03/tony-h…

Had a really fun conversation with @collision about the future of software engineering and Sierra over a couple of pints

Meta-benchmark idea: A collection of basic agent harnesses (AGENTS.md, docs, skills, hooks) that encapsulate non-trivial applications, each with a single prompt to terraform the application in one shot starting from scratch. When new coding agents or models come out, we can run against the harness to directly compare to previous model versions. “This model wrote a browser from scratch from the standard browser harness with X% feature completeness and only Y lines of code in Z hours for $XYZ.” Imagine seeing that browser Cursor developed through the lens of each subsequent model update. How good was the code? How fast was the renderer? How complete was the feature set? How long did the agent have to run? What was the all-in cost of the tokens? Kind of a CSS Zen Garden for agent harnesses, which did a ton to push forward browser standards. I feel like all of us have experienced the difference in Codex 5.4 Extra High (if not, stop reading this post and try it immediately), but we still don’t have a way of capturing in a benchmark the qualitative experience of using it as a developer, particularly for measures like architecture and code quality.

GPT-5.4 Thinking and GPT-5.4 Pro are rolling out now in ChatGPT. GPT-5.4 is also now available in the API and Codex. GPT-5.4 brings our advances in reasoning, coding, and agentic workflows into one frontier model.

I enjoyed sharing the stage with Ng Tian Chong at Mobile World Congress to discuss how Sierra partners with Singtel and our shared vision for driving real business impact with AI

I’ve been trying to simulate using Codex for the next year and what will change about my perspectives on software engineering as I transition from being a computer programmer to a harness engineer. There are so many, but here are a couple that have stuck with me: Software dependencies - Large open source systems like Linux and MySQL seem like they will remain just as important, but I wonder if I will start to have different perspectives on smaller software libraries when the functionality can be relatively easily produced and tested with AI. Given the past decade of supply chain vulnerabilities and maintenance issues in open source libraries, will it become a best practice to reduce dependencies and write our own where possible? Documentation - When I built a product before, the “specification” was split between docs, Slack, Figma, and Linear — but the vast majority of behavior was specified in code, i.e., the long tail of functionality is an emergent property of the code I write. The conundrum with agent-produced code is that it’s not clear which parts of the code were prompted (i.e., specified) and which parts were “vibed” (i.e., unspecified). That seems problematic when continuously evolving a large system over time because the harness will “forget” past instructions. I don’t think replaying prompts is correct either because in a single Codex session, a good chunk of interactions are interactive and effectively transient. I have an intuition that documentation will be as important of an output of my Codex sessions as code, documenting the substantive product decisions made during my session. Those docs clearly need to be directly in the repo, versioned with the code and available as context for future sessions. The docs / context discussion in OpenAI’s recent post on harness engineering resonated with me and maps to my intuition: openai.com/index/harness-…

Hear from the Next team directly about their experience working with Sierra

I am so proud of Sierra’s partnership with Next, the iconic fashion retailer. This is a story of speed and scale: we went live in just 6 weeks, and the agent is live in 83 countries! sierra.ai/customers/next

Excited to join Devesh Raj from Sky in a fireside chat at Mobile World Congress to discuss how Sky is leveraging Sierra to drive growth and delivering more human, connected customer experiences with AI. Request a spot here docs.google.com/forms/d/e/1FAI…

Love that Alysa Liu is from my hometown of Oakland ❤️🇺🇸

Had a great conversation with @jaltma about SaaSpocalypse (SaaSmageddon?), Sierra, and the future of software

This week, on Benchmark's new podcast Uncapped 😂, I sat down with @btaylor, founder of Sierra and Chairman of OpenAI. He's easily one of the most impressive people I’ve met in tech or in general. We talked about AI and the saaspocalypse, the unique considerations of building an AI native / agent company, whether young or experienced founders have the advantage right now, Codex and OpenAI ads, and much more. Learned a ton from Bret, hope you enjoy. (0:00) Intro (0:20) The Saaspocalypse and systems of record (12:34) Sierra's landscape (17:05) Outcome-based pricing (24:22) The rapid evolution of AI support technology (28:21) Young founders vs. experienced founders (34:12) What comes next beyond support (38:47) Codex and the future of software engineering (51:49) OpenAI and advertising (54:59) Working with investors and boards

When a patient calls 📞 about a confusing bill, speed matters — but so does how that moment is handled. Learn more from Cedar and Sierra at ViVE on Monday, February 23 at 4:45PM. hlth.com/events/vive/ag…

Excited to join Tian Chong Ng from Singtel at Mobile World Congress to discuss how Sierra has partnered with Singtel to build AI agents that drive real business growth

At MWC Barcelona, @btaylor and @Singtel CEO Tian Chong Ng will take the stage together on March 4 at 12PM for a keynote: "Building AI agents that drive real business growth". If you'll be at MWC, come find us. Learn more about the keynote: lnkd.in/gdsdezBW