The AI Record
what we were realizing is if there's a person that's deciding yes or no every time, at some point, the person actually just stops reading, and they just say yes, yes, yes, because they're tired of pressing yes. … And our security people were also realizing this. And so … the thing that they started working on is, is there a way to route this to a classifier? So the classifier can decide. … It took many months to kind of get this to a mature state. And in the end, through EVALS, through red teaming, through pen testing, we're able to show that this is actually much safer. And so … at Anthropic, we run on Auto Mode, and we recommend this to all our customers, too.

Boris ChernyCreator and head of Claude Code, Anthropic · Leader, Anthropic

Context. Cherny is explaining why Anthropic's security team worked on routing permission decisions to a classifier.

Podcast: Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster), AMD (YouTube), July 29, 2026, at 34:33.

Checked against the source [ Grade C ] Said in the last year Anthropic: $47 billion revenue, run-rate, May 2026 Sells AI tools

Filed under

Who is speaking

Boris Cherny, Creator and head of Claude Code. Leads the product and describes how Anthropic's own engineers use it, so he is in a position to know the internal practice. Anthropic sells Claude Code, so his account is first-party and promotional, and the figures he gives are spoken claims.

Anthropic sells AI or agent tools or services, so it has a reason to publish flattering numbers about AI. The passage stays in the archive; read it with that in mind.

How this was checked

Checked against the source by a first reader, then confirmed independently by a second reader who fetched the source without seeing the first reader's notes (Independent review agent (Claude Sonnet), blind second pass from passage text and citation only, no repository access, 2026-11-06).

Near 34:33. Whisper matches exactly ('the thing that they started working on', 'EVALS', 'Auto Mode'); the captions drop 'that' in 'the thing that they started'. The ellipses cover fillers and the omitted line about noticing it himself.

How to cite this passage

Boris Cherny (Anthropic), “Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster)”, AMD (YouTube), 2026-07-29, at 34:33. https://www.youtube.com/watch?v=eQ6tb7j3Z2U Archived at https://the-ai-record.com/passages/anthropic-run-on-auto-mode-classifier-instead-of-yes-2026

More from the same source

And … since the beginning of this year at Anthropic, we've seen an 8X increase in code output per engineer. And this is just unheard of in the industry. … And the thing that we did to get to the … 8X is we systematically improve a bottleneck at a time. … We throw [Claude] at it and [Claude] is the coding. The next bottleneck is the code review. … And so just like one step at a time, we unbottleneck every part of the process.

Boris ChernyCreator and head of Claude Code, Anthropic · Leader, Anthropic

Context. Cherny is describing what he says Anthropic has seen in engineering output this year and how the company got there.

Podcast: Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster), AMD (YouTube), July 29, 2026, at 14:09.

  • 8 x code output per engineer, since the beginning of the year (2026), as spoken. Level not stated

Checked against the source [ Grade C ] Said in the last year Anthropic: $47 billion revenue, run-rate, May 2026 Sells AI tools Passage page

At Anthropic, [Claude] is at the center of everything that we do. It's every single business process, every single bit of everything everyone does throughout the day. [Claude] is just at the center of it. … And so, … [Claude] writes the code. It does the code review. It does the security review. It brainstorms and comes up with new product ideas. It aggregates user feedback. It triages incidents. … And … now this is happening not just for engineering. It's happening for product and for design and for marketing and for GTM, for every function.

Boris ChernyCreator and head of Claude Code, Anthropic · Leader, Anthropic

Context. Cherny is answering the host's question about what his typical day at Anthropic looks like.

Podcast: Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster), AMD (YouTube), July 29, 2026, at 11:46.

Checked against the source [ Grade C ] Said in the last year Anthropic: $47 billion revenue, run-rate, May 2026 Sells AI tools Passage page

And honestly, [Claude Code] didn't work very well for … the first six months. … And it wasn't until, I would say, Opus 4 in May of 2025, where it really started to work. And … since then, the growth for … the product has been exponential also. It was not at first, but that's when it became exponential.

Boris ChernyCreator and head of Claude Code, Anthropic · Leader, Anthropic

Context. Cherny is explaining why Anthropic built a purely agentic product and how it developed.

Podcast: Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster), AMD (YouTube), July 29, 2026, at 8:05.

Checked against the source [ Grade C ] Said in the last year Anthropic: $47 billion revenue, run-rate, May 2026 Sells AI tools Passage page

And the way we get there is with dynamic workflows. And this is essentially using [Claude] to orchestrate just very large teams of agents to do the most complex work. Like, large … migrations. … Like, we just migrated … from Zig to Rust. And this is like a JavaScript runtime. It's a lot of code.

Boris ChernyCreator and head of Claude Code, Anthropic · Leader, Anthropic

Context. Cherny is describing what he calls dynamic workflows, in which the model coordinates many agents.

Podcast: Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster), AMD (YouTube), July 29, 2026, at 21:44.

Checked against the source [ Grade C ] Said in the last year Anthropic: $47 billion revenue, run-rate, May 2026 Sells AI tools Passage page

But really, for a lot of people, it's actually just really exciting to get to try all these tools to get to experiment. And so I actually see a lot of my role as creating space for people to try stuff out, to feel safe doing it. … And my other job is to just give people context, so to give them business context, product context, so that they can make better decisions.

Boris ChernyCreator and head of Claude Code, Anthropic · Leader, Anthropic

Context. Cherny is answering the host's remark that learning cycles are vastly accelerated, about how leaders should take people with them.

Podcast: Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster), AMD (YouTube), July 29, 2026, at 18:41.

Checked against the source [ Grade C ] Said in the last year Anthropic: $47 billion revenue, run-rate, May 2026 Sells AI tools Passage page

And I think … when I look at most companies, there's still somewhere between step one and two. … And when I look at Anthropic, we're probably somewhere around step three on average. So most engineers are running dozens or hundreds of agents. And the way you do this is you have the agent run another agent. And we have a lot of tools for this.

Boris ChernyCreator and head of Claude Code, Anthropic · Leader, Anthropic

Context. Cherny has just described stages companies go through, from every engineer running one agent to 10, 100 and then 1,000.

Podcast: Agentic AI and the Future of Software Development: S3 E4 (AMD Advanced Insights, hosted by Mark Papermaster), AMD (YouTube), July 29, 2026, at 20:38.

Checked against the source [ Grade C ] Said in the last year Anthropic: $47 billion revenue, run-rate, May 2026 Sells AI tools Passage page

Something wrong? Report a correction. JSON: /api/v1/passages/anthropic-run-on-auto-mode-classifier-instead-of-yes-2026.json