The hosts open with Anthropic’s Claude Opus 4.7 release, discussing Mythos, higher token usage, stronger visual understanding, and what a more agentic model means in practice. From there, they move into Anthropic’s growing tension with government access, speculation about a Figma competitor, and OpenAI’s push to make Codex a broader desktop and workflow tool. The middle of the episode focuses on Google’s AI mode, Gemini desktop possibilities, and how browser control and computer use could reshape product design. In the second half, they pivot to Google’s Disco, Luma’s virtual filmmaking workflow, Perplexity Personal Computer, Salesforce going headless for agents, and Allbirds’ strange compute pivot. Key Points Discussed 00:01:33 Claude Opus 4.7 and Mythos 00:08:56 White House Access to Mythos 00:12:12 Anthropic, Figma, and AI Design Tools 00:18:34 OpenAI Codex for Everything 00:24:41 Google AI Mode and Gemini Desktop 00:37:17 Google Disco and Agentic Research 00:40:38 Luma, Wonder Project, and AI Filmmaking 00:51:07 Perplexity Personal Computer 00:59:47 Salesforce Headless and the Agent-First Web 01:03:39 Allbirds Pivots to Compute The Daily AI Show Co Hosts: Karl Yeh, Andy Halliday, Beth Lyons, Brian Maucere

Can Agents Replace the Web?

The Daily AI Show

# Full Transcript

**Podcast:** The Daily AI Show

## Transcript

Hey, what's going on everybody? Welcome to The Daily AI Show. Today is April 17th, 2026. Appreciate you being here. This is episode 705 and with me today are Andy, Beth, and I'm Brian. Happy to be here. Happy to be back after being away for two days. I know you guys have been holding out on the fort as well as some other folks hanging in as well. So I know I sort of, because of editing, know some of the stories. It looked like yesterday I actually went back to the AI when I was extracting.

the AI stories and I got, I went back to ChatGPT, the custom GPT I have, and I was like, no, give me all the AI news stories. And it's like, you're right, there was one more and it went from two to three. But man, they must have really had some long conversations yesterday and not talked about AI dude because it only brought up three, which is unusual for us. So I've been out of the loop a little bit, but you guys have been here, Andy. I know you've been here all the days. So thank you Cisco. He's saying welcome back. So why don't,

Andy, why don't you kick us off? You would have the best, the best transfer knowledge of where we are with the news and we'll kind of go from there.

Yeah. And there's a, I have eight, there's eight real stories that we can talk about here, but I guess the lead one is what we announced all live on our show yesterday is the release of Claude Opus 4.7, which is showing smatterings of the improvements that are, you know, in the development pipeline of mythos, which, you know, in, in, as part of the announcement,

we showed yesterday on, on the screen, the comparison all across the benchmarks for Opus 4.7 against the other players. And also they published at the same time, the performance of mythos, which was way ahead of everybody else.

And is reinforcing this idea that Anthropic does have this pretty super powerful model behind the scenes. It's not been released to the general public yet, and probably maybe might not be.

It could just be used by cybersecurity people to, you know, defend existing systems, which take a long time to harden against this kind of capability that may be just used for that purposes, you know, teaming against, you know, the increasing security developments there.

But let's talk about 4.7 because it is an improvement and its pricing is unchanged from 4.6, but the notes on it showed that it can use up to 35% more tokens for identical text preparation and results.

And also if you use the extra high mode for additional thinking, that's going to burn up your context faster. And so it's, it's not a totally efficient model and it's not intended to be. This is their frontier, you know, publicly available frontier reasoning model. Yeah, you might, you might have a good point about that.

As I, as I look at cloud code, I've been in cloud code all morning and it was defaulted to Opus 4.7. So I didn't touch it. But I've used 51% of my context window. O...

Can Agents Replace the Web?

Transcript

Built for investment professionals

Every day you wait, your competitors are finding
edge in conversations you're not hearing.

Can Agents Replace the Web?

Transcript

Built for investment professionals

Every day you wait, your competitors are findingedge in conversations you're not hearing.

Every day you wait, your competitors are finding
edge in conversations you're not hearing.