Bret Taylor: belief

Transcript context

…It feels to me like a very interesting time in agentic engineering because you were talking about this domain of harness engineering and people having skills and MCP and everything like that. It's always interesting when not only is the leading product in a category changing, we're actually just figuring out what the categories are that we need things for MCP or skills or stuff like that, and it's all very fast-moving. That just feels to me like a very interesting time where clearly a new way of engineering is shaking out. 2026 is clearly not the final word. Absolutely. In fact, I'm growing more skeptical of MCP as a meaningful part of the future. It's fine as a protocol, but it's interesting. Going back to your joke around OpenClaw just writing a big markdown file. I think it works better than a bunch of MCP servers. But going back to the point of every AI agent knows how to use grep and knows how to use all these things, I feel like this view of a multi-agent world was you have all these agents that do tasks that are fraud detection, another one over here for personalization. Then you make a super agent, it does all these things, and it looks really good on a whiteboard like most elegant-looking but completely nonsensical architectures. Then you realize if you just imagine you anthropomorphize the Stripe experience and you're the checkout concierge. What information do you need to have no a priori to actually make that a humane experience? What ends up happening in these multi-agent systems is you stuff all the context in the subagents, and the one on top has no ability to actually not sound robotic. Then in contrast, you look at something like OpenClaw, it's just a bunch of markdown files, and the memory feels right, even though it's a little bit kludgey. Similarly, if you go back to my arguments about a source control, like a repo, it has so much context. It's not like you just have the myopic view of the file you're editing. It really has some expansiveness. My sense is we're making true agents over time. The way we think about context and how that context is shared so that the agent that's orchestrating it actually understands what's behind all these APIs and why and the history will maybe look a little bit more like OpenClaw and less like MCP over time. I think these agents need a lot more context than what MCP affords. One thing we've noticed... There's a bit of a what's old is new again phenomenon where with this agentic commerce stuff that's happening. We actually built the APIs for this 10 years ago as part of... Do you remember that move of social shopping that was in for a while? Buying on Instagram, buying on Twitter. It didn't quite happen for a few different reasons at the time. But the concepts are very similar that you want some action at a distance, you want to be able to go manipulate stuff off-site. Similarly, I think Patrick has wanted for the longest time in Stripe is the ability to just SSH into your Stripe account. What do you mean? It's a very ergonomic way for developers to work is you should just be able to log into your Stripe account and you have a command line there, and then you can list out all your... Or you can tail the payments log.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.