Back in 2024, Google CEO Sundar Pichai defended AI Overviews amid concerns they could reduce traffic to websites: “The summary we are providing clearly adds value.” Look out for our latest interview with Google’s CEO later this week on Decoder.

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Our guest today is The Verge’s Hayden Field, and we’re discussing a question that’s been rocketing around the tech industry for the past week: Is Google losing the AI race? That's because last week, the company announced a bombshell reorganization of its AI division. You can read these moves, and Google’s reaction, in a lot of different ways. So we really wanted to sit down with Hayden to dig into some of the smartest analysis we’ve seen this past week, and what we think is really going on here. #google #ai #deepmind #technology

0:00 - Cold Open: Is Google losing the AI race?
0:44 - Welcome to Decoder
0:57 - Google's AI reorganization: Jeff Dean & Demis Hassabis departures
2:38 - Google's advantages but falling behind
3:31 - Sundar Pichai on reorganizing for AI
5:25 - Gemini 4 and frontier lab status
6:10 - SemiAnalysis: "DeepMind is no longer a frontier lab"
7:39 - Can Google reach state of the art again?
9:10 - Google's bureaucratic culture problem
10:30 - Brain drain and moral concerns at Google
12:58 - Funding AI with revenue vs. values
15:58 - Ad break (Engine)
18:59 - Does Google want to win at the frontier or sell cloud?
21:14 - AGI, the singularity, and Demis's vision
24:47 - Product vs. research tension
27:33 - DeepMind's London autonomy changing
28:58 - Koray Kavukcuoglu takes over DeepMind
30:56 - Can Google actually lose the AI race?
33:38 - Signs to watch for from Google
35:08 - Outro

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

For our latest episode of Decoder, we sat down with Allan Thygesen, the CEO of Docusign. You know Docusign, it’s the platform that lets you sign stuff online. It turns out 7,000 people work there, which is one of those facts floating around that’s always felt like perfect Decoder bait. What are all those people doing? And what kind of product roadmap does a company like Docusign even need?

Allan has only been CEO of Docusign for three years, so he has some interesting perspective on where the company was, the changes he wanted to make, and where he thinks this is all going. Hint: it involves AI. #ai #docusign #legal #tech

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

The biggest story in tech right now is the spiraling debate about AI safety and regulation. The CEOs of all these companies have issued various calls to slow down development and develop regulation, including asking for antitrust exemptions so they can all coordinate on safety issues.

So to make sense of it all, I called Jonathan Kanter, the former antitrust chief for the U.S. Department of Justice in the Biden Administration. There's a lot going on, and we really got into the weeds of how all this regulation might work, and how it might interact with antitrust law, competition policy, and our relationship with China.

0:00 Cold Open
0:46 Welcome to Decoder
1:41 Introducing Jonathan Kanter
2:39 The State of AI Regulation
3:58 "Rules of the Road" for AI
4:39 AI companies & Antitrust Exemption
8:53 "Robot Jail" & Product Liability for AI
11:18 Congress Is Broken
13:46 Section 230 & Public Trust
16:44 The Strange Political Realignment
19:09 Little Tech vs. Big Tech
21:01 Decoder Sessions
22:33 China as a Convenient Boogeyman
25:33 Competition Law vs. Foreign Policy
29:00 Should We Break Up the AI Oligopoly?
33:15 The Political Scramble Before Midterms
35:11 End of the Reagan Era & the New Deal
36:38 The Supreme Court Dismantled Regulation
38:05 Hipster antitrust
42:19 Why Trump Hasn't Settled the Apple Case
43:39 Kanter's Prescriptions for AI Regulation
Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Our guest today is Lila Snyder, the CEO of Bose. Lila was last on the show about four years ago, and since then she’s dramatically rearchitected the company — she sold off Bose Professional, acquired the high end audio brands McIntosh Labs and Sonus faber, and, perhaps most importantly, started licensing all that core Bose technology to other companies.That big change in the business comes with a big change in structure. Becoming a software company in addition to a product company comes with a whole new set of challenges, and Lila got into how it’s all working. #bose #audio #headphones #tech

0:00 – Teaser
0:36 – Intro & history
2:44 – Multi-brand: McIntosh & Sonus Faber
3:04 – B2B audio tech licensing
5:31 – "Powered by Bose" explained
8:50 – Protecting luxury brands from being "Bose-ified"
13:31 – Is luxury just expensive versions of the same thing?
17:53 – Modular R&D = faster innovation
19:11 – How Bose is structured today
22:14 – Are software algorithms protectable?
24:14 – Commodification risk
28:00 – Decision-making framework
29:40 – AI wearables & third-party players
34:00 – Tech always sacrifices audio quality for convenience
37:54 – AI platform shift: Meta, Apple, natural language input
40:55 – Recording everything; audio as AI hardware opportunity
43:19 – Noise-canceling hearing aids (Bose + Orka)
46:15 – AR glasses, hearing aids, headphones
49:29 – Tariffs & supply chains
51:07 – Bluetooth: still the limiting factor
54:10 – Home speakers are now network computers
57:33 – Smart home AI agents will commodify audio further
58:33 – Bose Studios: content & creator community
1:00:27 – Closing

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

In this 2024 clip, Google CEO Sundar Pichai takes a look at a sample Google search on Nilay’s phone to evaluate whether search results with AI will be better or worse than they were before. Look out for our latest interview with Google’s CEO next week on Decoder.

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Dr. Jill Lepore is one of the best writers there is at identifying institutional patterns in history, and Decoder is a show about systems and organization — so her new book, The Rise and Fall of the Artificial State, is a perfect fit for the show.

You build systems by measuring things — gathering information, quantifying it as data, and taking action on it. The book's big thesis is that this kind of systemic measurement is really a dehumanizing act, one that flattens humanity and human concerns into something, well, artificial.

These artificial structures touch everything — not just in the more obvious places like in apps or AI, but all the way up to politics, governance, and the very future of the nation-state itself.

0:00 - Cold open
0:54 - Intro
2:27 - Interview starts
3:18 - Defining "the artificial state"
5:15 - Public discourse privatized, bots dominate platforms
6:39 - The trajectory toward losing liberal democracy
7:49 - 1930s: the invention of scientific polling
8:47 - How data mining replaced door-to-door campaigning
9:20 - AI's role in 2026 election messaging
10:36 - Sci-fi's dystopian visions vs. Silicon Valley's embrace of them
11:36 - Asimov's "Franchise" and Sam Altman on AI choosing presidents
12:19 - Silicon Valley's rejection of humanism and the past
14:23 - Uber advertising segment with Christy Argilon
16:41 - Peter Thiel, Elon Musk, and religion as moral cover
18:27 - Setting aside "the mess" of democracy for efficiency
20:14 - The 1980s and the Macintosh's "1984" ad
22:39 - The personal computer and internet as "just an ad campaign"
24:26 - Nilay pushes back: the internet did democratize access
27:27 - How platforms consolidated power (Musk, Zuckerberg)
29:28 - Newspapers, historical parallels to media democratization
31:20 - Seven party systems and the cycle of elites regaining control
32:37 - Identity politics and algorithmic sorting
34:35 - 1968 predictions about personal newspapers and political fragmentation
38:08 - Targeted advertising and the modern ad economy
41:31 - Consumer culture replacing citizenship
43:52 - Living in an internet built for bots, not humans
46:30 - Tech CEOs asked what happens when bots replace users
48:15 - Apple's App Store fees and the "bots only" internet fantasy
50:19 - College grads booing AI - a moment of pushback
52:23 - Yale's AI task force story
48:38 - Students demand real discussion about AI policy
47:56 - Data centers and local democratic pushback
54:30 - Democracy reasserting itself


Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

My basic reaction to the AI regulation debate is that I feel like everyone has forgotten why we even have governments in the first place #ai #tech #politics

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Our guest today is Shishir Mehrotra, the CEO of Superhuman, the company formerly known as Grammarly, which is still its flagship product. Back in August, Grammarly shipped a feature called Expert Review, which allowed you to get writing suggestions from AI-cloned “experts,” and recently, reporters at The Verge and other outlets discovered that those experts included Nilay, among many others.

No one ever asked permission to use these people’s names this way, and a lot of reporters were outraged by this. To Shishir’s credit, he did not cancel our interview and he came on and stuck it out. This conversation got tense at times, and it’s clear there’s disagreement about how extractive AI feels for people. There’s a lot in this one, and we’re excited to hear what you think. #ai #superhuman #grammarly #creators #tech

Read the full interview transcript on The Verge: http://bit.ly/47SBKIT

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

We’re back with another Decoder mailbag special. This time, we focused on feedback to Nilay’s “software brain” video essay, our ongoing coverage of AI and Big Tech backlash, surveillance, interviewing politicians ahead of the midterms, and more. Remember: we really do read every email, so please keep sending us your feedback, and letting us know what you think of our coverage and how we can improve. #ai #tech #surveillance

0:00 Cold Open: AI Surveillance & UBI
0:41 Show Intro & Welcome to Decoder
1:29 The Mailbag Episode
2:56 The Origin of Nilay's Rants
4:42 Verifiability & the Limits of AI
7:26 Reader Responses to Software Brain
9:00 Natural Language vs. Traditional Interfaces
12:27 Sam Altman on Delivering Value
13:46 Control, Agency, & Meaning
14:48 Cloudflare CEO Preview: AI in the Workplace
15:56 Surveillance & Flock Camera Backlash
20:54 Sundar Pichai & Google Zero
26:11 PMC/PMX Disclosure
27:03 China's AI Adoption vs. Western Skepticism
31:44 UBI & Sam Altman
32:09 Algorithmic Fragmentation & Tech Backlash
37:10 Governor Hochul Interview Feedback
40:49 Bluesky CEO & The Liberal Bubble
45:18 How Companies Pressure The Verge
49:30 Outro & Credits

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Hard to start an AI regulator when we’re in the middle of unplugging the government and plugging it back in again, I think

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Our guest today is Ryan Mac, a technology reporter at The New York Times and co-author of the excellent book Character Limit: How Elon Musk Destroyed Twitter, which came out in 2024. We wanted to have Ryan on the show because we’re on the cusp of the SpaceX IPO, which promises to be one of the most consequential public offerings in history for a variety of reasons.

Its biggest-ever size, of course, at nearly $2 trillion dollars. But also because all kinds of rules that keep our markets fair are being bent, if not outright broken, along the way. And, also because buried somewhere inside SpaceX is X, the social platform formerly known as Twitter, which Musk purchased in 2022. #elonmusk #spacex #xai #ai #tech

0:00 Cold open
0:43 Introduction
2:14 Interview begins
2:42 SpaceX IPO overview
4:10 X (Twitter) is stagnating
5:37 Musk's net worth & path to trillionaire
9:42 Tesla reputational damage vs. SpaceX/Starlink
12:27 Why users haven't left X
16:52 Corporate governance — Elon's 85% voting control
21:14 Index funds — everyone ends up owning SpaceX
22:29 Nasdaq-100 rules relaxed for SpaceX
28:41 SpaceX fundamentals — Starlink is the only profit center
31:47 $28 trillion TAM & AI ambitions
36:58 xAI/Grok — "the RC Cola" of AI
39:13 Closing — accountability & along for the ride
42:41 Outro

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Cloudflare discovered in June that bots already make up more than half of internet traffic — and CEO Matthew Prince says that bots could outnumber humans a thousand to one within five years. That's a huge change to the internet, and if the web is going to keep existing, it needs a huge change to its business model. Can Cloudflare use its place in the ecosystem to direct the money back to the people who make content? #cloudflare #web #google #ai #tech

0:00 - Cold Open
0:43 - Show Intro
2:35 - What Is Cloudflare?
4:51 - The Business Model Of The Internet Is Changing
10:32 - Web As Info Platform Vs. App Platform
14:05 - The Spotify Analogy For Information Markets
23:30 - Local News Becomes Valuable Again
26:39 - Blocking AI Crawlers And Google
36:40 - Distribution Changes What You Make
39:12 - The Attention Economy Is Broken
41:28 - Content Moderation And Cloudflare's Power
53:22 - Mission And Values
54:13 - The AI Layoffs: Builders, Sellers, Measurers
1:01:09 - AI Excels At Measurement
1:06:40 - AI Identifying Rising Stars
1:11:28 - Will AI Dehumanize Workers?
1:16:09 - Show Outro

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Our guest today is Robert Hart, The Verge’s London-based AI reporter. Robert recently wrote a fantastic story for us about the debate raging inside the world of mathematics — and the existential crisis over what it means that new frontier AI models have become very good at math in a shockingly short period of time. So we wanted to dive into all of this with Robert, who actually spoke to some of the most accomplished mathematicians working today to figure out what’s hype and what’s real — and to get a sense of just how exciting, and how scary, all of this is. #math #ai #openai #technology

0:00 Intro
1:43 Interview: Robert Hart on AI & math
3:20 Why AI got better at math
7:13 OpenAI's Astra & 10 breakthroughs
9:02 Attribution & credit controversy
12:09 Repeatability
14:12 Logic, code & verifiability
15:36 How did the models do this?
17:25 Mowing down problems vs. progress
19:58 The PhD crisis
21:52 Gary Marcus & skepticism
24:38 Democratizing math
27:38 The Leiden Declaration
30:57 Economic incentives & pure research
33:27 Optimism vs. shell shock
34:50 Outro

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Our guest today is Ben Smith, the editor-in-chief of Semafor. Ben ran BuzzFeed News at the height of the Facebook traffic boom, and even wrote a book about it, called Traffic, which he discussed with me on Decoder back in 2023.

Semafor, which is only four years old, is now a fast-growing and profitable media company, which makes about half of its revenue from events. So we asked Ben how that all works, and what it feels like to navigate the ethical conflicts that seem to be everywhere in media right now. #media #news #tech #journalism

0:00 Cold Open & Intro
2:36 Interview begins: How Semafor Is Doing
3:28 Lessons from BuzzFeed & the Traffic Era
6:00 Reacting to Jonah Peretti on Facebook & BuzzFeed
10:04 YouTube, Creators & the Anti-Scale Model
13:12 The Convening Business & Why Semafor Stays Free
16:07 Expansion, the Semaform & Semafor's Structure
25:54 Journalism Ethics & Independence
29:17 Silicon Valley & The World: Advisory Boards
36:08 News Creators, Access & the Clip Economy
42:06 Advertising, Platforms & the Coming Reckoning
53:11 Politics, Influence Ops & AI Regulation
59:34 Gulf States, Free Speech & AI in the Newsroom
1:06:13 Wrap Up

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Our guest today is Hayden Field, senior AI reporter for The Verge. Often when Hayden comes on the show, it’s because something has gone wrong in the world of AI. Last weekend, that something was a pretty intense mix of Anthropic, the Trump administration, and Anthropic’s new AI model, Fable 5.

Hayden actually just published a fantastic play-by-play on The Verge about how the Fable ban went down, and the scramble through the weekend from both sides to figure out what exactly happened and how it might get resolved. So we wanted her to come on and just walk us through the timeline and what it all means. #tech #ai #anthropic #claude

0:00 Intro
0:52 Welcome to Decoder
3:36 What is Claude Mythos & Fable 5?
6:41 The jailbreak & 90-minute ultimatum
14:42 Why did Andy Jassy kneecap his own investment?
16:04 Why only Anthropic?
25:37 Was wiping out the Biden framework a mistake?
27:27 What would real AI regulation look like?
32:18 Should AI be treated like military tech?
34:55 What's next?

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Today, I’m talking with Mustafa Suleyman, the CEO of Microsoft AI. As you’re no doubt aware, the biggest story in tech right now is the spiraling debate about AI safety and regulation. So I really wanted to talk to Mustafa about what he thinks is real and not in AI safety, whether the concept of alignment itself is up to the task, and whether this industry needs to slow down before it kills us all. #microsoft #ai #aisafety #regulation #anthropic

0:00 Cold Open
1:01 Introducing Mustafa Suleyman
2:04 Interview Starts - Is Alignment Broken?
7:06 The Hugging Face Incident
10:03 Imposing Industry Standards
12:46 Why Publish the Memo Now
15:09 Anthropic & Model Welfare
17:48 Government's Rejection of Regulation
21:16 Product Liability & Historical Regulation
24:38 What's Lost on Twitter
26:34 Deep Dive: Anthropic's Claude Constitution
35:31 The China Race Framing
39:04 The Public Trust Gap
41:33 Superintelligence vs. AGI
43:35 Calling for a Slowdown
46:39 Regulating Open Models Locally
48:11 What's Next / Closing

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Today I’m talking with Josh Dzieza, a longtime features writer here at The Verge, about a massive data center in Utah that does not exist, might never exist, and yet has already completely reshaped the politics of the entire state. Josh just published a major feature this week on the struggling Stratos data center project in Utah backed by ‘Shark Tank’ star Kevin O’Leary. So I invited Josh on the show to discuss what’s happening inside the AI industry’s infrastructure boom, inside the bipartisan backlash to data centers, and how the tech industry is running into the one obstacle it can’t seem to spend its way out of: local democracy. #ai #datacenters #tech

Decoder is a Signal Awards finalist in the Business category! You can vote to help us win here: https://bit.ly/47wsyJv

📷: Stephen Voss, Getty Images

0:00 - Cold Open
0:52 - Show Intro
1:57 - The Wonder Valley Story
5:28 - How O'Leary Got Approvals
8:12 - The Community Finds Out
13:05 - Who Are the Protesters?
15:39 - How the Opposition Succeeded
18:26 - Kevin O'Leary's Downfall
21:19 - The Data Center Race for Speed
26:14 - Does AI Economics Matter?
28:43 - What If the AI Bubble Pops?
31:12 - Lessons for the Rest of the Country
33:12 - Trump & Data Centers
37:05 - NIMBYism vs. Real Concerns
38:13 - What Happens Next?
40:04 - Outro

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Cloudflare CEO Matthew Prince on why he felt had to do layoffs amid the AI boom: "If we're real leaders, we should be the ones taking the arrows."

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

My guest today is longtime friend of the show Joanna Stern. You all know Joanna: she is the former senior personal technology columnist for The Wall Street Journal, a former Decoder guest host, one of my co-founders at The Verge, and also just one of my very closest friends.

Joanna just left that lofty perch at the Journal to start her own media company called New Things, and she’s starting with her new book about AI called I Am Not a Robot, which is out this week on May 12th. So we had Joanna on to talk about all of that, especially what she learned going all in on automation.

0:00 Intro / Cold open
0:54 Welcome & introducing Joanna Stern
2:01 Joanna's new book "I Am Not a Robot" & New Things
3:13 Are consumer AI products actually good?
7:15 Is there a killer consumer AI product?
13:23 What AI has Joanna kept using after the experiment?
16:41 Wearable AI: Meta glasses & the Bee Bracelet
21:32 Culture change & humanoid robots
27:20 Why AI companies ship half-baked products
31:34 Privacy trade-offs: surveillance, cost vs. convenience
33:40 AI companions & Joanna's AI boyfriend
40:14 Rules for AI & no phones for kids
43:00 Starting New Things: why YouTube?
49:09 Medium is the message: algorithm vs. editorial judgment
52:07 Data narrows you / CarPlay teaser

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Connecting with Google CEO Sundar Pichai at I/O every year is one of our favorite Decoder traditions. This was our fifth year doing it, and there’s always a whole slew of new things to talk about. This year, in addition to the news, we talked about Google Zero; picking fights with YouTube creators and publishers; and what being at “the foothills of the singularity" even means. #google #ai #agi #tech

0:00 – Intro
0:41 – Introduction & episode overview
3:44 – How Google is structured today
8:53 – How Sundar restructured Google in response to ChatGPT
11:55 – Decision-making framework
15:23 – Could AI replace the CEO?
17:06 – AI agents, Gemini Spark & Antigravity explained
22:55 – Live search demo: AI Overview vs. organic results
26:51 – Public anxiety around AI — can Google close the gap?
33:21 – Google Zero & the health of the open web
33:53 – Condé Nast CEO says "assume zero search traffic"
37:56 – Should publishers & YouTube creators opt out of AI training?
41:57 – "Foothills of the singularity" — what does it mean?
43:28 – Can LLMs get us to AGI?
45:21 – Sundar's AGI timeline

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Adam Conover is a familiar face to anyone who’s spent time on the internet. He got his start at CollegeHumor back in the day, before transitioning into cable TV with Adam Ruins Everything, then onto Netflix, and all the way back to YouTube again, where he posts videos and hosts a podcast. That is exactly the kind of career path that fascinates me, and it also happens to touch on basically every major Decoder theme there is.

0:00 — Cold Open
0:41 — Welcome to Decoder
1:47 — Introducing Adam Conover
3:48 — "Creator" Is a Trick Word
5:11 — Tech Giants vs. Hollywood
7:03 — YouTube Is the New TV
9:44 — The Death of Mid-Budget TV
11:23 — The Old Cable Business Model
14:27 — Forced to Be an Entrepreneur
16:05 — Rebuilding Comedy Infrastructure
18:14 — How Adam's Business Works
20:30 — The YouTube Race to the Bottom
22:19 — The Future of YouTube
24:11 — How Adam Makes Decisions
27:36 — The Pressure to Go Viral
32:39 — Algorithm vs. Creative Instinct
36:20 — YouTube Is the Gatekeeper
39:58 — Brand Integrations & Advertising
45:00 — The Worldcoin Ad Controversy
50:01 — Union organizing YouTubers
54:39 — Government Regulation & Labor
55:48 — Hollywood's Future
59:39 — Movies as America's Cultural Legacy
1:00:13 — Outro

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Our guest today is Hank Green, a longtime friend of Decoder and the co-founder and now former owner of Complexly, an online education company he started with his brother John in 2012. We say former owner because Hank and John have just converted Complexly into a nonprofit and given up their ownership of the company in the process.

That’s some of the purest Decoder bait that ever was, because it’s absolutely all about organization and decisions and how you choose to make that kind of big, structural change to a company. So of course we had to bring Hank back on to talk about it. #youtube #hankgreen #education #tech #ai

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

Mr Wonderful does not think our data center reporting is very wonderful at all

Follow Decoder on TikTok: @decoderpod
Follow Decoder on Instagram: @decoderpod

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

It’s a big week for Meta. The company has its big Connect conference, and the new Muse AI agent appears to be an early hit. CEO Mark Zuckerberg has been doing a lot of interviews about it all — so of course we asked for him to be on Decoder. Meta declined, for whatever reason. But luckily Mark just did a big interview with our friend and former colleague Alex Heath on the Sources podcast. So I figured we could highlight some moments from that conversation to point out some obvious gaps in what Mark is saying about Meta and hopefully inspire the next set of people who talk to him with some follow-up questions. #ai #meta #muse #technology

Sources with Alex Heath: Mark Zuckerberg on how Meta's Muse AI agent will make people money: https://www.youtube.com/watch?v=Lx8lrn-cytc

0:00 - Intro
0:20 - Reacting To Mark Zuckerberg's Sources Podcast Interview
0:45 - Meta Glasses: Passive Surveillance Debate
3:21 - Mark's Response On Glasses Communication
6:09 - Teen Safety & Meta's State Settlement
8:21 - Meta Gets Punished For Transparency
10:45 - Mark's Go-It-Alone Approach To AI Safety
12:19 - Muse Business Model & Aggregation Theory
14:46 - Data Center Investments
17:17 - Free Muse Usage & "A Lot Of Computer"
17:52 - Closing

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

My guest today is Mike Cannon-Brookes, who is co-founder and CEO of Atlassian. Atlassian is one of those companies that every other company runs on — it makes important platform tools like Jira and Trello. This means Atlassian is also right in the middle of the way AI is changing how all these companies work. Or, if you buy the idea of the so-called Sasspocalpyse, AI might just build all of these tools for you, destroying this entire category of businesses. Obviously, Mike had a lot of thoughts pushing back on this narrative, so we spent time really digging in and talking through what AI is doing to enterprise software. #ai #atlassian #software #technology

0:00 Cold Open
0:41 Welcome To Decoder
1:03 What Is Atlassian Today
3:26 How AI Changes Work
9:51 The SaaSpocalypse Debate
13:44 How Atlassian Is Structured
16:28 Running On Its Own Platform
21:48 Platform-Centric Decision Making
25:16 Layoffs & Skill Mix Changes
29:25 Role Blending In The AI Era
34:50 Org Charts Of The Future
37:58 Supply Vs Demand Constrained Jobs
39:42 Imagination Asymmetry
41:25 How Mike Makes Decisions
44:31 Headless Is Brainless
46:45 MCP Servers & Connected Platform
53:11 The Dia Browser
1:05:20 Consumer AI Products
1:10:04 What's Next For Atlassian
1:11:00 Wrap Up

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

Link to Resources: https://thenextnewthing.ai/my-resources
Presented by Zapier: https://zapier.com/
Andrew Warner and Tashfeen Ahmed reveal how to get free LLM access for AI coding, agents, image generation, video, audio, and more using Free LLM API.

Andrew sits down with Tashfeen Ahmed, creator of Free LLM API, to show how developers can access free token quotas from multiple AI providers through a single unified API. Tashfeen demonstrates how the platform automatically routes requests across providers, supports automatic failover, and lets users prioritize reliability, speed, or intelligence when choosing models. They walk through the setup process on Mac, Windows, Docker, iOS, and Android, showing how to connect the unified API key to coding agents like Claude Code and OpenCode and start experimenting without paying for tokens.

They also explain why free LLM access is particularly useful for prototyping and experimentation, while cautioning against relying on provider free quotas in production applications when the provider doesn't permit it. Tashfeen shows how Free LLM API uses Thompson Sampling to continuously evaluate models and route requests, and how the platform supports more than just text models, including embeddings, image, video, audio, and multi-model Fusion. The episode is a practical look at stretching free AI access further while experimenting with different models and providers.

Links featured:

- Free LLM API — https://github.com/tashfeenahmed/freellmapi
- Zapier SDK — https://zapier.com/sdk

00:00 - Free LLM API
00:27 - $500+ in Free AI Tokens
01:03 - Models, Providers & Automatic Failover
02:06 - Keeping the Same Model Across Providers
02:42 - How to Get Started
03:36 - 45 AI Providers With One API
03:54 - Using Free LLM API With AI Coding Agents
04:30 - Can Free LLM Access Get You in Trouble?
05:24 - Why AI Providers Offer Free Quotas
05:42 - Free Models in Production Apps
06:18 - Zapier SDK
06:36 - Smart Model Routing
07:12 - Thompson Sampling & Model Rankings
07:48 - Embeddings, Images, Video & Audio
08:15 - Free LLM API Desktop App
08:42 - More Ways to Minimize Token Usage

Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Presented by Zapier: https://zapier.com/

👉 Link to resources: https://thenextnewthing.ai/l/evan-mcp-tools
👉 Evan Nison (LinkedIn): https://www.linkedin.com/in/evan-nison/

What if your AI could actually use your computer for you?

In this episode of The Next New Thing, Andrew Warner talks with Evan Nison, who built tools that let AI agents control your browser, send messages, and complete real tasks across your desktop—all from the terminal.

Evan demos how his Browser Bridge and Desktop Bridge connect tools like Claude Code to real-world actions: filling out forms, navigating websites, sending iMessages, and even testing front-end code automatically. Instead of guessing from code or screenshots, the AI can now see, click, and fix things on its own.

They also break down how this fits into a broader shift toward MCP servers and agent workflows—where APIs, automation tools like Zapier, and direct computer control all work together.

The bigger idea: AI isn’t just generating outputs anymore—it’s becoming an operator, capable of executing real work across your entire system.

⏱ Timestamps
00:00 - AI controlling your browser from the terminal
00:18 - What Browser Bridge actually does
00:45 - Using MCP servers with Zapier and browser control
01:21 - Why APIs are still faster than browser actions
01:48 - Real use case filling out forms automatically
02:33 - Watching AI control the browser live
03:18 - How Evan structures data for automation
03:54 - Why human review still matters before submission
04:12 - Desktop Bridge controlling your computer
04:48 - Sending iMessages with AI
05:24 - Replacing manual outreach with automation
06:00 - How the system sends messages instantly
06:27 - Why Evan trusts Claude over other agents
07:03 - Bigger picture your own “OpenClaw” setup
07:30 - Use cases for browser and desktop control
08:06 - Why AI that can self-check is better
08:42 - Building Browser Bridge with Claude Code
09:18 - Iterating until the AI tools work together
09:54 - 180 plus tools inside Browser Bridge
10:30 - Using AI for front-end QA and debugging
10:57 - Automating authentication and setup
11:24 - Using Browser Bridge to build its own website
12:00 - Why it improves design and branding
12:27 - How Evan actually uses Zapier agents
13:03 - Biz dev prep agent example
13:30 - Email bounce recovery automation
13:57 - Why Zapier is still the easiest layer
14:24 - Final thoughts and real-world use cases

👉 Join us: https://thenextnewthing.ai/

Today, I’m talking with Bloomberg’s Mark Gurman, who is the world’s best-sourced Apple reporter. Earlier this month, Apple hosted its annual iPhone hardware event, and this event was special, for a number of reasons. It was new Apple CEO John Ternus’ first major product launch. It was also the debut of the foldable iPhone Duo. So I wanted Mark on the show to explain why this event was so important, and how John Ternus’ Apple will look a lot different from Tim Cook’s. #apple #iphone #iphoneduo #siri #ai

0:00 Cold Open: Apple's Privacy Problem
0:58 Welcome to Decoder
2:38 Covering Apple as a Fan
3:34 Apple's Community Across Eras
4:28 Ternus vs. Cook: Management Styles
5:40 The Intelligent Personal Hub Story
8:14 Apple AI at the Event
9:34 AI-Native Device Roadmap
10:56 Jobs Era vs. Cook Era
13:06 Why Cook Let Ternus Take the Stage
15:00 The Succession Plan
17:07 "Hotel California" Strategy
19:33 Phil Schiller and the App Store
22:23 Org Structure Under Ternus
23:29 Johnny Srouji and Apple Silicon
28:18 The Cook Doctrine and AI Strategy
31:08 Jony Ive, OpenAI, and the Lawsuit
38:17 AirPods with Cameras
40:20 Apple's Privacy Challenge
47:10 Meta Glasses and the Privacy Blind Spot
49:34 The iPhone Duo
51:21 Ternus' Challenges and Opportunities
53:59 Outro

Subscribe: http://goo.gl/G5RXGs
Like The Verge on Facebook: https://goo.gl/2P1aGc
Follow on Twitter: https://goo.gl/XTWX61
Follow on Instagram: https://goo.gl/7ZeLvX
Follow on TikTok: http://bit.ly/42VeCVU

Watch The Vergecast on YouTube: https://bit.ly/40RFRkg
The Vergecast Podcast: https://bit.ly/3WQDexZ
Decoder with Nilay Patel: http://apple.co/3v29nDc
More about our podcasts: https://www.theverge.com/podcasts

Read More: http://www.theverge.com
Community guidelines: http://bit.ly/2D0hlAv
Wallpapers from The Verge: https://bit.ly/2xQXYJr
Shop our Verge merch store here: https://bit.ly/4kPCmEc

If you buy something from a Verge link, Vox Media may receive a commission without exerting any influence on editorial content. For more information about our ethics policy, visit: https://bit.ly/3ZWTlLs

My review of The Witcher 3 — Remastered!

Twitter: https://twitter.com/NeonKnightYT

Charlamagne Tha God, co-host of “The Breakfast Club,” talks with Anderson about grief and his experiences with anxiety and depression. He reflects on the death of his close friend Jas Waters, healing from childhood trauma and why grief is something “you can’t run from.”

0:00 The emotion that Charlamagne Tha God is most afraid of feeling
3:05 Charlamagne's mental health struggles
10:01 How Charlamagne's experience with childhood molestation has affected his life
17:28 Charlamagne explains how ayahuasca has helped him process his grief
21:50 The importance of checking in on your friends' mental health

For more of “All There Is with Anderson Cooper” visit https://cnn.com/allthereis

#AllThereIs #Grief #CNN

Bloomberg Pursuits reporter Sarah Rappaport tells Bloomberg This Weekend that luxury travelers are increasingly seeking tiny hotels with as few as one to six rooms, combining the privacy of a villa with personalized high-end hotel service. Speaking with hosts David Gura and Christina Ruffini, Rappaport says the trend is spreading across Europe and the US as travelers seek exclusivity, fewer crowds and unusual experiences in restaurants, museums and historic properties.
--------
More on Bloomberg Television and Markets

Like this video? Subscribe and turn on notifications so you don't miss any videos from Bloomberg Markets & Finance: https://tinyurl.com/ysu5b8a9
Visit http://www.bloomberg.com for business news & analysis, up-to-the-minute market data, features, profiles and more.

Connect with Bloomberg Television on:
X: https://twitter.com/BloombergTV
Facebook: https://www.facebook.com/BloombergTelevision
Instagram: https://www.instagram.com/bloombergtv/

Connect with Bloomberg Business on:
X: https://twitter.com/business
Facebook: https://www.facebook.com/bloombergbusiness
Instagram: https://www.instagram.com/bloombergbusiness/
TikTok: https://www.tiktok.com/@bloombergbusiness?lang=en
Reddit: https://www.reddit.com/r/bloomberg/
LinkedIn: https://www.linkedin.com/company/bloomberg-news/

More from Bloomberg:
Bloomberg Radio: https://twitter.com/BloombergRadio

Bloomberg Surveillance: https://twitter.com/bsurveillance
Bloomberg Politics: https://twitter.com/bpolitics
Bloomberg Originals: https://twitter.com/bbgoriginals

Watch more on YouTube:
Bloomberg Technology: https://www.youtube.com/@BloombergTechnology
Bloomberg Originals: https://www.youtube.com/@business
Bloomberg Quicktake: https://www.youtube.com/@BloombergQuicktake
Bloomberg Espanol: https://www.youtube.com/@bloomberg_espanol
Bloomberg Podcasts: https://www.youtube.com/@BloombergPodcasts

This week, National Economic Council Director Kevin Hassett says AI’s productivity boost may be much larger than official economic data currently indicate. And, AI companies are largely in agreement that the technology needs guardrails, but the ethical implications of AI could also create new opportunities for liberal arts education. Plus, America is rebuilding a critical minerals supply chain after decades of Chinese dominance. Later, America’s shrinking college-age population is forcing universities to compete for fewer students.

Chapters:
00:00:58 - Productivity vs. Deficits
00:11:22 - Thinking With AI
00:23:44 - Rare Earths Race
00:35:39 - College Enrollment Cliff
00:47:20 - How AI Reshapes Work
--------
More on Bloomberg Television and Markets

Like this video? Subscribe and turn on notifications so you don't miss any videos from Bloomberg Markets & Finance: https://tinyurl.com/ysu5b8a9
Visit http://www.bloomberg.com for business news & analysis, up-to-the-minute market data, features, profiles and more.

Connect with Bloomberg Television on:
X: https://twitter.com/BloombergTV
Facebook: https://www.facebook.com/BloombergTelevision
Instagram: https://www.instagram.com/bloombergtv/

Connect with Bloomberg Business on:
X: https://twitter.com/business
Facebook: https://www.facebook.com/bloombergbusiness
Instagram: https://www.instagram.com/bloombergbusiness/
TikTok: https://www.tiktok.com/@bloombergbusiness?lang=en
Reddit: https://www.reddit.com/r/bloomberg/
LinkedIn: https://www.linkedin.com/company/bloomberg-news/

More from Bloomberg:
Bloomberg Radio: https://twitter.com/BloombergRadio

Bloomberg Surveillance: https://twitter.com/bsurveillance
Bloomberg Politics: https://twitter.com/bpolitics
Bloomberg Originals: https://twitter.com/bbgoriginals

Watch more on YouTube:
Bloomberg Technology: https://www.youtube.com/@BloombergTechnology
Bloomberg Originals: https://www.youtube.com/@business
Bloomberg Quicktake: https://www.youtube.com/@BloombergQuicktake
Bloomberg Espanol: https://www.youtube.com/@bloomberg_espanol
Bloomberg Podcasts: https://www.youtube.com/@BloombergPodcasts

iTunes Download: http://msclvr.co/38tHqR Google Play: http://smarturl.it/dg-nbt
Stream from Spotify/Apple/etc: http://smarturl.it/s-nbt

It was a long hike to get to this place... much longer than we bargained for :1 Hope you guy like this video for "Next Best Thing" off Sam's album "Make It Up"

For Physical CD, T-shirts and more, click here:
http://www.kurthugoschneider.com/store
_______________________________

GET IN TOUCH!

SAM TSUI:
Facebook: http://www.facebook.com/samtsuimusic
Twitter: http://www.twitter.com/samueltsui
YouTube: http://www.youtube.com/thesamtsui

KURT SCHNEIDER:
Facebook: http://www.facebook.com/kurthugoschneider
Twitter: http://www.twitter.com/kurthschneider

"Frozen 2" is available on Disney+

Together with Anna, Kristoff, Olaf and Sven, Elsa faces a dangerous but remarkable journey into the unknown—to the enchanted forests and dark seas beyond Arendelle, in search of truths about the past.

🌟 Every Magical Moment, Every Iconic Song- on DisneyMusicVEVO ➜ https://www.youtube.com/@Disne....yMusicVEVO?sub_confi

All Your Favorite Disney Hits, All Together: https://www.youtube.com/playli....st?list=PLiNVoBckLqL

Are you still repeatedly eating egg sandwiches every time you visit Japan?

Come on, let’s try the next big trend already.

In today’s video, I’m going to show you 8 delicious 7-Eleven items recommended by local Japanese people.
They’re all so good that I wouldn’t be surprised if one of these becomes the next big hit.
Trust me, you have to try them.


Music
Music by Nicknames - Summertime Fine - https://thmatc.co/?l=7EC5195E

Music by KORALOO - BREAKUPWITHYOURSELF - https://thmatc.co/?l=037320E1

#seveneleven
#japanesefood
#japantravel
#conveniencestore

What did Mojang announce at Minecraft Live in September 2026? Oh, not much. Just that The Sift is coming to Minecraft Java and Bedrock as the first new dimension in over a decade; Minecraft Dungeons II releases September 29th, 2026; the Holiday Drop adds Ice Caves and a frozen zombie variant; Minecraft World theme park opens in 2027 at Chessington World of Adventures Resort in London; and the new Creator Platform lets you create, manage, and share community content. Minecraft also launches on Nintendo Switch 2 on October 27th. Phew! Did we miss anything? Go watch the entire show to find out!

Watch Minecraft LIVE: https://youtube.com/live/Is5IswM87EQ

American Sign Language: https://youtube.com/live/-uQFmnbPZPo
British Sign Language: https://youtube.com/live/bj_Z0XAsuvE
Audio Description: https://youtube.com/live/Buh2UiDT7kM

Get Minecraft Dungeons II at: https://www.minecraft.net/stor....e/minecraft-dungeons

Check out Minecraft.net for more information!
https://www.minecraft.net/

Presented by Zapier: https://zapier.com/

👉 Download the free report: https://thenextnewthing.ai/l/github-repos-april17
👉 Adam Brakhane (LinkedIn): https://www.linkedin.com/in/adam-brakhane/

New GitHub repos are changing how AI actually works—memory, voice, agents, and even full replacements for paid tools.

In this episode of The Next New Thing, Andrew Warner breaks down the most important repos of the week with Adam Brakhane—and what they mean for builders.

They cover tools that convert messy data into clean formats for LLMs, give Claude persistent memory, add voice to any workflow for free, and package entire agent stacks into a single subscription. Some are early. Some are rough. But each one points to a bigger shift in how AI systems are being built.

⏱ Timestamps
00:00 Intro
00:18 MarkItDown
00:54 Claude Memo
04:12 Kronos
06:09 VoxCPM
10:03 Karpathy Skills
11:06 Hermes Agent
14:06 Multica
16:03 Postiz
19:03 Gary Tan’s AI workflow
25:03 Final thoughts and full report link

👉 Join us: https://thenextnewthing.ai/

Resources: https://thenextnewthing.ai/my-resources
Presented by Zapier: https://zapier.com/
Andrew Warner and Michael Galpert compare Instinct, GrokBot, and Muse using the same real-world task.

Andrew Warner and Michael Galpert test Instinct, GrokBot, and Muse head-to-head using the same Airbnb search to see how each agent handles research, follow-ups, browser control, purchasing, speed, and usability. They also compare pricing, connectors, reusable agent personas, multiplayer coordination, and where each tool fits best for personal and business workflows.

Links featured:

Muse — https://www.muse.ai/
Grok Bot — https://x.ai/bot
Instinct — https://instinct.com/
Zapier MCP — https://zapier.com/mcp
Hyperagent — https://www.hyperagent.com/
Instinct invitation (thanks Rick Segal!) - https://app.instinct.com/invit....e?t=4wclizn6blahks5w

00:00 - Instinct vs GrokBot vs Muse: Test the same prompt across three new AI agents and compare how they respond.
00:18 - Instinct: Search for an Airbnb through iMessage and get recommendations without opening another app.
03:00 - GrokBot: Run the same Airbnb search with browser access, screenshots, maps, and clickable results.
05:24 - Muse: Compare listings in a cleaner interface with ratings, cancellation dates, and fast follow-up questions.
07:48 - Agent Speed: See how response time changes the way you interact with Muse, Instinct, and GrokBot.
09:09 - Live Muse Test: Search for an Airbnb in Mexico City and watch Muse browse listings in real time.
11:06 - Zapier MCP: Connect agents to 8,000+ apps while restricting what they can access and change.
11:51 - Pricing: Compare Muse’s free tier, Instinct’s free access, and GrokBot’s paid usage.
12:18 - Connectors & Personas: Build custom connections in Muse and reusable skills, processes, and agent personas in GrokBot.
13:39 - Instinct Privacy & Trust: See the tradeoff between its simple iMessage interface and less visibility into what the agent is doing.
14:42 - Multiplayer Agents: Let separate Instinct agents coordinate schedules, trips, and shared preferences between people.
17:42 - Where Each Agent Fits: Compare GrokBot for complex agent workflows, Instinct for ease of use, and Muse for the Meta ecosystem.
19:03 - Hyperagent: Andrew explains why he prefers Hyperagent for company and team use.

👉 Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Resources: https://thenextnewthing.ai/l/astra-reaction-sep12
Presented by Zapier: https://zapier.com/
Andrew Warner breaks down the best real-world GPT-6 Astra use cases and how it compares with Fable 5.1.

Andrew Warner tests where GPT-6 Astra actually stands out by reviewing real-world demos from Riley Brown, Jason Lee, Claire Vo, Matthew Berman, Mark Kashef, Creator Magic, Nate Herk, and Dan Shipper. The episode covers app building, browser and computer control, iPhone automation, DaVinci Resolve editing through MCP, meeting analysis, writing, and presentation creation, while comparing Astra with Fable 5.1 on quality, speed, cost, and practical usefulness.

Links featured:

Riley Brown — https://www.youtube.com/watch?v=Ju41cQSe7hY&t=489s
Jason Lee — https://www.youtube.com/watch?v=hjAugywVkOw&t=1180s
Claire Vo (How I AI) — https://www.youtube.com/watch?v=AniiF8rOu9c&t=350s
Matthew Berman — https://www.youtube.com/watch?v=xdXLzFzxA9Q&t=706s
Mark Kashef — https://www.youtube.com/watch?v=tU-fO6cADvQ&t=836s
Creator Magic — https://www.youtube.com/watch?v=V7CtHz8JFWA&t=510s
Nate Herk — https://www.youtube.com/watch?v=WfJPBVXPt8k&t=865s
Dan Shipper (Every) — https://www.youtube.com/watch?v=1EEw36H2zLo&t=73s

00:00 - GPT-6 Astra: Andrew breaks down the best real-world use cases he found after testing Astra and watching hours of demos.
00:27 - Raycast-Style App: Build a custom desktop utility from a simple prompt in under 20 minutes.
01:39 - $60K/Month App Clone: Replicate a receipt-scanning bookkeeping app using GPT-6 Astra and Fable 5.1.
02:51 - GPT-6 Astra App Build: Scan receipts, extract data with Claude, and generate a polished working app from one prompt.
04:12 - Fable 5.1 App Build: Compare Fable’s version of the same receipt-scanning app against Astra.
06:18 - Computer Use: Let GPT-6 Astra take over Chrome and modify a complex browser-based workflow hands-free.
07:39 - Zapier MCP: Connect AI agents to thousands of apps while controlling what they can access and change.
08:42 - Browser Control: Use GPT-6 Astra to draw in Excalidraw, research products, and plan routes in Google Maps.
10:12 - iPhone Control: Control an iPhone through Mac’s iPhone Mirroring using GPT-6 Astra.
12:00 - DaVinci Resolve MCP: Connect GPT-6 Astra directly to DaVinci Resolve for faster AI-powered video editing.
15:18 - AI Video Editing: Turn more than 150GB of event footage into a finished recap with GPT-6 Astra and Fable 5.1.
19:12 - Meeting Analysis: Analyze dozens of meetings to uncover business problems and automation opportunities.
22:12 - AI Writing: See why Dan Shipper considers GPT-6 Astra a strong writing model.
24:09 - Consulting Decks: Compare GPT-6 Astra and Fable 5.1 creating a McKinsey-style presentation from the same prompt.
27:27 - GPT-6 Astra vs Fable 5.1: Andrew gives his overall verdict on computer use, writing, design, speed, and cost.

👉 Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

OpenCut hit 57K stars as a real CapCut alternative, Apple just shipped a native container tool for Apple Silicon, and a Netflix engineer found a trick to cut your AI agent token costs by 60%.

Presented by Zapier: https://zapier.com/
👉 Link to resources: https://thenextnewthing.ai/l/github-weekly-june19

Andrew Warner and Adam Brakhane break down 13 repos — from AI agent skills and PM workflows to a system prompt leak that reveals how Claude Fable 5 actually thinks, and a security scanner that catches malware hiding inside skills.

Timestamps

00:00 - Intro
00:36 - Hidden Gems
00:45 - coreyhaines31/marketingskills - AI marketing workflows for SEO, social, ads, and onboarding.
02:15 - mattpocock/skills - Personal skills for engineering, learning, and productivity.
05:06 - OpenCut-app/OpenCut - An open-source CapCut alternative you can run yourself.
06:54 - apple/container - Apple’s native tool for running Linux containers on Mac.
09:09 - NVIDIA/SkillSpector - Scans AI agent skills for risky behavior.
12:27 - addyosmani/agent-skills - Helps coding agents follow a stronger engineering workflow.
14:51 - Zapier MCP - Connect AI tools to apps through one controlled toolbox.
15:45 - phuryn/pm-skills - Product management skills for clearer requirements and positioning.
18:00 - iptv-org/iptv - Public IPTV channels playable through tools like VLC.
19:39 - Panniantong/Agent-Reach - Helps agents access sites that block simple scraping.
21:00 - chopratejas/headroom - Reduces unnecessary token usage between tools and models.
24:27 - mvanhorn/last30days-skill - Fresh research from GitHub, YouTube, Hacker News, X, and Polymarket.
27:00 - music-assistant/server - Connects smart-home audio with music services and local sources.
28:57 - asgeirtj/system_prompts_leaks - System prompts from major AI products for builders to study.
33:36 - Watch this next

We explore the TOP 10 trending repos this week.
Full repo report: https://thenextnewthing.ai/l/github-repos-jul3
Presented by Zapier: https://zapier.com/
Peter's YouTube channel: https://www.youtube.com/@cooperx86

Andrew Warner and Peter Cooper break down the top trending GitHub repos of the week, starting with xbtlin/ai-berkshire, a Claude Code investment research framework that analyzes companies through the lens of Warren Buffett, Charlie Munger, Duan Yongping, and Li Lu.

They also cover SimpleX Chat, OpenMontage, google-labs-code/design.md, codebase-memory-mcp, no-mistakes, ai-website-cloner-template, lingbot-map, free-for-dev, and Orca. Along the way, they look at a few useful honorable mentions, including ponytail, Nextcloud Server, and Inbox Zero.

Chapters:

00:00 - Intro
00:27 - xbtlin/ai-berkshire: Turn Claude Code into a value-investing research team.
02:24 - simplex-chat/simplex-chat: A private messenger with no phone number, username, or user ID.
04:03 - Help Andrew get Matt Van Horn on the show
04:39 - calesthio/OpenMontage: Generate videos by combining agent workflows, stock footage, narration, music, and rendering tools.
06:27 - Zapier MCP: Connect agents to Gmail, Calendar, Drive, Notion, and thousands of other tools.
08:06 - DietrichGebert/ponytail: Make AI coding agents solve tasks with the simplest code possible.
09:54 - nextcloud/server: Self-host your own alternative to Google Drive, Calendar, Contacts, and more.
10:39 - elie222/inbox-zero: A self-hostable AI email assistant for organizing and drafting replies.
12:09 - google-labs-code/design.md: Give AI agents a reusable design system so outputs match your brand.
15:09 - DeusData/codebase-memory-mcp: Give coding agents codebase memory and reduce wasted token usage.
16:30 - kunchenguid/no-mistakes: Add an AI safety gate before code reaches your team’s repo.
17:51 - JCodesMore/ai-website-cloner-template: Rebuild an existing site as a clean modern codebase.
20:51 - Robbyant/lingbot-map: Reconstruct a 3D scene from streaming camera frames.
22:48 - ripienaar/free-for-dev: A huge maintained list of free developer tools and service tiers.
25:21 - stablyai/orca: Run multiple AI coding agents in parallel from one desktop environment.
27:09 - Watch last week’s GitHub repo breakdown next

👉 Join us: https://thenextnewthing.ai/

Resources: https://thenextnewthing.ai/l/astra-reaction-sep12
Presented by Zapier: https://zapier.com/
Andrew Warner breaks down the best real-world GPT-6 Astra use cases and how it compares with Fable 5.1.

Andrew Warner tests where GPT-6 Astra actually stands out by reviewing real-world demos from Riley Brown, Jason Lee, Claire Vo, Matthew Berman, Mark Kashef, Creator Magic, Nate Herk, and Dan Shipper. The episode covers app building, browser and computer control, iPhone automation, DaVinci Resolve editing through MCP, meeting analysis, writing, and presentation creation, while comparing Astra with Fable 5.1 on quality, speed, cost, and practical usefulness.

Links featured:

Riley Brown — https://www.youtube.com/watch?v=Ju41cQSe7hY&t=489s
Jason Lee — https://www.youtube.com/watch?v=hjAugywVkOw&t=1180s
Claire Vo (How I AI) — https://www.youtube.com/watch?v=AniiF8rOu9c&t=350s
Matthew Berman — https://www.youtube.com/watch?v=xdXLzFzxA9Q&t=706s
Mark Kashef — https://www.youtube.com/watch?v=tU-fO6cADvQ&t=836s
Creator Magic — https://www.youtube.com/watch?v=V7CtHz8JFWA&t=510s
Nate Herk — https://www.youtube.com/watch?v=WfJPBVXPt8k&t=865s
Dan Shipper (Every) — https://www.youtube.com/watch?v=1EEw36H2zLo&t=73s

00:00 - GPT-6 Astra: Andrew breaks down the best real-world use cases he found after testing Astra and watching hours of demos.
00:27 - Raycast-Style App: Build a custom desktop utility from a simple prompt in under 20 minutes.
01:39 - $60K/Month App Clone: Replicate a receipt-scanning bookkeeping app using GPT-6 Astra and Fable 5.1.
02:51 - GPT-6 Astra App Build: Scan receipts, extract data with Claude, and generate a polished working app from one prompt.
04:12 - Fable 5.1 App Build: Compare Fable’s version of the same receipt-scanning app against Astra.
06:18 - Computer Use: Let GPT-6 Astra take over Chrome and modify a complex browser-based workflow hands-free.
07:39 - Zapier MCP: Connect AI agents to thousands of apps while controlling what they can access and change.
08:42 - Browser Control: Use GPT-6 Astra to draw in Excalidraw, research products, and plan routes in Google Maps.
10:12 - iPhone Control: Control an iPhone through Mac’s iPhone Mirroring using GPT-6 Astra.
12:00 - DaVinci Resolve MCP: Connect GPT-6 Astra directly to DaVinci Resolve for faster AI-powered video editing.
15:18 - AI Video Editing: Turn more than 150GB of event footage into a finished recap with GPT-6 Astra and Fable 5.1.
19:12 - Meeting Analysis: Analyze dozens of meetings to uncover business problems and automation opportunities.
22:12 - AI Writing: See why Dan Shipper considers GPT-6 Astra a strong writing model.
24:09 - Consulting Decks: Compare GPT-6 Astra and Fable 5.1 creating a McKinsey-style presentation from the same prompt.
27:27 - GPT-6 Astra vs Fable 5.1: Andrew gives his overall verdict on computer use, writing, design, speed, and cost.

👉 Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Check out Muse, your personal AI agent. Redeem my code in Settings within 48 hours of joining and we'll both get 1 billion Muse tokens.

Code: IF17B3
Link: https://muse.ai/join

Link to Resources: https://thenextnewthing.ai/my-resources
Presented by Zapier: https://zapier.com/
Get your private invite or say hi to Michael: https://x.com/msg
Andrew Warner and Michael Galpert take an inside look at Muse, Meta’s new AI agent with its own computer, and explore what it can actually do across Instagram, Facebook, WhatsApp, web browsing, and everyday workflows.

Andrew sits down with Michael Galpert, an early OpenClaw user, to test Muse and see how far its agent capabilities go. Michael demonstrates how Muse can access Instagram data, export information into an Excel file, visualize relationships between followers, manage messages, and work with Meta’s ecosystem in ways that other AI agents often cannot. They explore Muse’s human-in-the-loop approval system, persistent memory, identity and personality settings, side chats, connectors, and its ability to communicate through channels like WhatsApp. Muse can also operate a cloud computer, open a browser, research products, and let users take control when needed.

The conversation also explores where Muse shines—and where it still falls short. Michael explains why the agent is particularly useful for people working with Facebook Marketplace, Facebook Ads, and Instagram, while Andrew and Michael test its ability to connect tools that don't have a built-in integration. They discuss Tailscale access to other computers, security and anti-spam protections, limitations with websites like Costco, and the difference between having one persistent agent versus multiple specialized agents. They also explore Muse’s personalized feeds and Ideas tab, which can suggest useful automations based on what the agent knows about you. After using Muse for several days, Michael shares why he sees it as a serious contender for how people may interact with AI agents in the future.

Links featured:

Muse — https://www.muse.ai/
OpenClaw — https://openclaw.ai/
Tailscale — https://tailscale.com/
Zapier SDK — https://zapier.com/sdk

00:00 - Introducing Muse
00:19 - Exporting Instagram Followers
01:38 - What Makes Muse Different
02:00 - Messaging on Facebook & Instagram
02:46 - Inbox, Memory & Agent Identity
03:25 - Human Approval for Messages
04:05 - Bulk Messaging
04:29 - Muse's Personality & Memory
05:11 - One Persistent Agent + Side Chats
06:12 - Muse's Connectors
07:34 - A Free Computer in the Cloud
07:53 - Using Muse Through WhatsApp
08:57 - Muse Controls a Web Browser
10:29 - Connecting Tools Without Built-In Integrations
11:20 - Creating Connectors on the Fly
12:51 - How Fast Is Muse?
13:18 - What Muse Is Good For
13:33 - Muse for Facebook & Instagram
14:27 - Security & Privacy
15:10 - Where Muse Still Falls Short
15:47 - One Agent vs. Multiple Personalities
16:17 - Trying Agents for Free
16:42 - Muse's Personalized Feeds
18:02 - Muse's Ideas Tab
18:47 - Michael's Muse Hot Take

Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Link to Resources: https://thenextnewthing.ai/my-resources
Presented by Zapier: https://zapier.com/
Zapier X Post: https://x.com/zapier/status/21....02829768332767453?s=
Andrew Warner sits down with Zapier co-founder and CEO Wade Foster to show how Next Gen Zaps can make AI agents more reliable, reduce unnecessary token costs, and prevent agents from making expensive or unexpected mistakes.

Andrew shares two real-world examples of the problem: spending $264 in just two days because an AI agent repeatedly checked websites, and an agent that replaced his own video thumbnails in sponsor emails with thumbnails from other creators because it decided they would perform better. Wade explains how Next Gen Zaps solve this by combining deterministic automation with agentic decision-making, so agents can handle the parts that require judgment while predictable actions are “hardened” to run the same way every time.

Wade demonstrates the approach using a Granola-to-Google Drive workflow. The system checks for new meeting transcripts, uses AI to determine whether a transcript contains sensitive topics that should not be automatically shared, and then deterministically saves approved transcripts to Google Drive. He also walks through a fantasy football workflow that researches rosters, waiver wires, injuries, and trade opportunities, uses AI for the managerial decisions, and then sends Wade a consistent weekly report.

The episode breaks down exactly where AI should be probabilistic and creative—and where it should be deterministic, predictable, and locked down. Wade also explains how agentically managed workflows can recover from errors automatically, giving you the flexibility of AI without making every step dependent on an agent.

Links featured:

- Zapier — https://zapier.com/
- Zapier Next Gen Zaps / Early Access — http://zapier.com/workflow/early-access
- Zapier X Post: https://x.com/zapier/status/21....02829768332767453?s=

Timestamps:

00:00 Why AI Agents Make Expensive Mistakes
01:30 Introducing Zapier Next Gen Zaps
01:39 Granola + Google Drive Demo
02:24 Adding AI to a Deterministic Workflow
03:09 Deciding What Should Stay Private
03:54 How Next Gen Zaps Work Under the Hood
04:57 Why Deterministic Workflows Matter
05:42 Agentically Managed Workflows
06:45 Why Hardened Automation Is Useful
08:15 Fantasy Football AI Workflow
09:27 Using AI to Analyze a Fantasy League
10:48 Turning the Analysis Into a Weekly Workflow
12:00 The Weekly Waiver & Trade Report
13:30 What Is Agentic vs. Deterministic?
14:15 Where AI Makes the Decisions
15:18 What Goes Wrong Without Deterministic Rules?
16:30 Why Hardened Workflows Matter
16:48 How to Get Early Access

Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Link to Resources: https://thenextnewthing.ai/l/matt-shownotes-sep17
Presented by Zapier: https://zapier.com/
Andrew Warner and Matt Van Horn break down the latest AI releases, including one-prompt native app building, Claude’s new Slides and Docs capabilities, AI wearables, WordPress agents, Notion skills, agent-friendly websites, and an AI receptionist that can answer and route business calls.

Andrew and Matt explore Modaal, which lets you build native iOS and Android apps from prompts, including Swift and Kotlin apps with native features like Live Activities and widgets. They then look at Claude’s new Slides and Docs capabilities and Andrew’s experiment turning a week of AI research into a presentation. Next up is YourPersona, an AI-powered wristband designed to learn your habits, answer questions, and proactively help with everyday tasks. They discuss whether a dedicated AI wearable can compete with devices people already wear, and why an agent-agnostic form factor could ultimately be more useful.

The episode also covers WPVibe, which brings AI-powered control to WordPress sites; Notion’s new skills library and what it could mean for sharing repeatable AI workflows with teams; Aside, which gives AI tools browser access; Aesty’s AI stylist; and Is Agentic, a Vercel tool that evaluates whether websites are accessible and usable by AI agents. Finally, Andrew and Matt explore ElevenLabs Reception, an AI receptionist that can answer calls, help customers, transfer conversations, and book appointments. Along the way, they discuss the shift toward agents that can interact with the real world, use personal context, and take action across the tools and services people already rely on.

Links featured:

- Modaal — https://modaal.dev/
- Claude Docs + Slides — https://claude.com/blog/cowork-is-now-claude/
- Persona Band — https://yourpersona.com/band
- WPVibe — https://wpvibe.ai/
- Notion Skills Library — https://www.notion.com/blog/a-....skills-library-for-e
- Aside — https://aside.com/
- Aesty — https://www.aesty.ai/
- Is Agentic — https://is-agentic.com/
- ElevenLabs Reception — https://elevenlabs.io/reception

00:00 - Modaal: Build Native Apps With One Prompt
01:30 - Native iOS & Android Apps
02:33 - Claude Slides & Docs
04:21 - YourPersona AI Wristband
05:15 - What Makes AI Wearables Interesting?
07:03 - The Case for Agent-Agnostic Wearables
07:30 - WPVibe: AI for WordPress
10:03 - Notion's Skills Library
11:24 - Why Skills Matter for Teams
14:15 - Aside: Give AI a Browser
15:36 - Aesty: AI Personal Stylist
16:48 - Is Agentic: Is Your Website AI-Friendly?
18:27 - Vercel Domains
20:42 - ElevenLabs Reception
22:39 - Muse Gets Phone Calls
23:42 - More AI Releases

Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Link to Resources: https://thenextnewthing.ai/l/c....orey-ganim-shownotes
Presented by Zapier: https://zapier.com/
Andrew Warner and Corey Ganim break down the latest AI releases, from Muse finding people ways to save money and Jev controlling a computer with voice, to AI-powered PR, polished product demos, ChatGPT Voice, vibe-coded games, webcam AI, digital commerce, and Claude Opus 5.5.

Muse, the free AI agent from Meta, and the growing number of real-world examples of people using it to find cheaper insurance, cancel forgotten subscriptions, find money owed to them, and more. They also look at Muse’s new voice capability and what could make it a gateway for everyday people to start using AI agents.

Next, explore Jev, a Mac app that lets you talk to your computer and have it take actions for you, and Cluso, which turns rough screen recordings into polished product demos, training videos, and SOPs. They also cover NewsJack, an open-source collection of agent skills that can turn an AI agent into a PR team by researching journalists, finding relevant stories, and generating pitches.

The episode also covers the latest ChatGPT Voice update, including access to email, calendars, Slack, and other connectors; a multiplayer tennis game built in just a few days with AI; Sider’s AI sidebar experience; Camera Bench, which lets AI interpret webcam input in real time; and a new digital mall built with Claude Opus 5.5. Finally, Andrew and Corey discuss why Opus 5.5 is standing out for writing, visual design, animation, and creative work.

Links featured:

ChatGPT Voice

- OpenAI announcement — https://x.com/OpenAI/status/2102808325742322002
- ChatGPT Voice Help — https://help.openai.com/en/art....icles/20001274-chatg

Muse

- Meta Muse — https://ai.meta.com/muse/
- Singularity Research — https://x.com/SingularityRes/s....tatus/20999661141052
- Muse — https://muse.ai/

Jev / TypeSafe AI

- TypeSafe AI — https://typesafe.ai/
- Jev demo — https://x.com/instantricecook/....status/2100814590300

Claude Opus 5.5

- Claude Opus 5.5 — https://www.anthropic.com/claude-opus-5-5
- Corey Ganim — https://x.com/coreyganim/statu....s/210279362460193211
- Other Reality — https://x.com/other__reality/s....tatus/21025423054337
- PDoomVideo — https://github.com/JohnHeibel/PDoomVideo

Systo Commerce

- Systo Commerce — https://systo-commerce.vercel.app/
- Systo AI — https://x.com/systoai/status/2103116401766752334

Glance Speedlab

- GitHub — https://github.com/yoheinakajima/glance-speedlab
- Yohai Benkler — https://x.com/yoheinakajima/st....atus/210286948147547

Top Tennis

- Top Tennis — https://toptennisgame.com/
- Demo — https://x.com/teovito/status/2102686586723033440

Sider Omni

- Sider Omni — https://sider.ai/lab/sider-omni

Clueso

- Clueso — https://www.clueso.io/

Newsjack

- GitHub — https://github.com/elvisun/newsjack
- Elvis Sun — https://x.com/elvissun/status/2100951347080421409

00:00 This Week’s AI Releases
00:19 Muse Is Saving People Money
01:09 AI Agents for Everyday People
01:56 Muse Gets Voice
02:11 Jev: Talk to Your Computer
03:09 CluesoTurns Rough Recordings Into Polished Demos
05:05 NewsJack: AI as Your PR Team
06:38 ChatGPT Voice Gets Connectors
08:34 Building a Game With AI
09:28 Sider Omni AI Sidebar
10:53 Camera Bench: AI Watching Your Webcam
12:35 A Walkable Digital Store
13:33 Claude Opus 5.5
14:17 Opus 5.5 for Animation & Creative Work
14:55 More AI Releases

Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Resources: https://thenextnewthing.ai/l/github-repos-sep18
Presented by Zapier: https://zapier.com/
This week, Andrew Warner and Adam break down 10 trending GitHub repos, from AI code security and context management to multi-agent workflows, knowledge graphs, and tools for running entire projects with AI.

They start with Alibaba’s Open Code Review, a tool that automatically checks code for security and other issues and can generate detailed HTML reports. Then they revisit I Have ADHD, which makes AI coding responses shorter and easier to scan, and God’s Eye View, which turns public data like traffic, cameras, and flight information into an interactive map.

The conversation also covers ECC’s engineering workflows and test-driven development, Context Mode’s approach to reducing context overhead, Tencent’s WeKnora for turning files into a searchable AI-powered knowledge graph, and OpenAI Plugins for extending coding agents with specialized tools and skills. They also explain Git worktrees and how Worktrunk lets multiple AI agents work on separate copies of the same project, plus Addy Osmani’s Agent Skills for a structured plan → build → test → review → ship workflow. Finally, they look at the Claude Code GitHub repo and why it exists even though Claude Code itself isn’t open source.

Repos featured:

- Alibaba Open Code Review — https://github.com/alibaba/open-code-review
- I Have ADHD — https://github.com/ayghri/i-have-adhd
- God’s Eye View — https://github.com/bilawalsidhu/gods-eye-view
- ECC — https://github.com/affaan-m/ECC
- Context Mode — https://github.com/mksglu/context-mode
- Tencent WeKnora — https://github.com/Tencent/WeKnora
- OpenAI Plugins — https://github.com/openai/plugins
- Worktrunk — https://github.com/max-sixty/worktrunk
- Agent Skills — https://github.com/addyosmani/agent-skills
- Claude Code — https://github.com/anthropics/claude-code

Audience submissions:

- Logic Loop — https://github.com/SuperLogicAI/Logic-Loop
- Claude Dashboard — https://github.com/JonImmsWord....pressDev/claude-dash
- Deck Gauge — https://github.com/Codpal-Limited/deckgauge
- Campus AIOS — https://github.com/trainingsites/campus-ai-os
- Reckoner — https://github.com/CaptainASIC/reckoner



Link Featured:

- Monid AI — https://monid.ai/?utm_source=creator&utm_content=newthinkg


Timestamps:

00:00 This Week’s GitHub Repos
00:28 Open Code Review by Alibaba
02:07 I Have ADHD
03:27 God’s Eye View
05:05 Monid AI
06:26 ECC
08:59 Should We Keep AI Skills?
10:17 Why Skills Matter for Less Capable Models
11:00 Context Mode
13:08 Tencent WeKnora
14:48 Zapier SDK & MCP
15:08 OpenAI Plugins
16:15 Plugin Security
17:00 Worktrunk & Git Worktrees
19:03 Agent Skills
21:04 Claude Code on GitHub
22:25 Audience Submissions
22:49 Logic Loop
23:32 Claude Dashboard
24:05 Deck Gauge
24:43 Campus AIOS
25:31 Reckoner
26:07 Wrap Up

👉 Send us your AI builds: hi@thenextnewthing.ai

Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Resources: https://thenextnewthing.ai/l/top-skills-sep16
Presented by Zapier: https://zapier.com/
Andrew Warner and Robbie Abed break down the 10 most-installed AI skills from GitHub, including skills that help agents find other skills, improve software design, browse the web, follow engineering best practices, and become better tutors.

Andrew and Robbie start with Find Skills, a Vercel skill that works like an app store for AI skills, helping agents discover and install the right capabilities for a task. They then explore Matt Pocock’s Grill Me and Grilling skills, which push AI to ask better questions and stress-test ideas before work begins. Anthropic’s Frontend Design skill is next, helping agents create more distinctive interfaces instead of falling into the familiar “AI-generated” design patterns. They also look at Vercel’s Agent Browser, which gives agents headless control of a web browser, and Matt Pocock’s setup skill for bringing engineering workflows and best practices into a repository.

The list also includes Vercel’s React Best Practices skill, which helps agents follow performance recommendations for React and Next.js, Lark Doc for interacting with ByteDance’s workplace platform, and Matt Pocock’s Teach skill, which turns an AI agent into a persistent tutor that builds lessons over time and remembers what you've already learned. Finally, they explore Vercel’s Web Design Guidelines, covering accessibility, forms, typography, animation, and other practical web standards. Throughout the episode, Andrew and Robbie show what these skills actually do, how they work together, and why packaging repeatable workflows into skills can make AI agents more capable.

Links featured:

- **find-skills** — vercel-labs/skills — (https://github.com/vercel-labs..../skills/blob/main/sk
- **grill-me** — mattpocock/skills — (https://github.com/mattpocock/....skills/blob/main/ski
- **frontend-design** — anthropics/skills — (https://github.com/anthropics/....skills/blob/main/ski
- **agent-browser** — vercel-labs/agent-browser — (https://github.com/vercel-labs..../agent-browser/blob/
- **setup-matt-pocock-skills** — mattpocock/skills — (https://github.com/mattpocock/....skills/blob/main/ski
- **vercel-react-best-practices** — vercel-labs/agent-skills — (https://github.com/vercel-labs..../agent-skills/blob/m
- **grilling** — mattpocock/skills — (https://github.com/mattpocock/....skills/blob/main/ski
- **lark-doc** — (https://github.com/larksuite/c....li/blob/main/skills/
- **teach** — mattpocock/skills — (https://github.com/mattpocock/....skills/blob/main/ski
- **web-design-guidelines** — vercel-labs/agent-skills — (https://github.com/vercel-labs..../agent-skills/blob/m

00:00 - Top 10 AI Skills
00:27 - Find Skills
01:03 - Grill Me
02:15 - Frontend Design
03:02 - Robbie Abed & NoMiddle
03:54 - Agent Browser
05:33 - Setup Matt Pocock Skills
06:27 - Vercel React Best Practices
07:39 - Grilling
08:42 - Lark Doc
09:36 - Teach
10:48 - Web Design Guidelines
11:24 - Why AI Skills Are Easy to Use
12:18 - Deep Dive With Matt Pocock

👉 Send us your AI builds: hi@thenextnewthing.ai

Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/

Resources: https://thenextnewthing.ai/my-resources
Presented by Zapier: https://zapier.com/
Andrew Warner and Henry Ndubuaku show how Needle brings small, private AI models directly into software and devices.

Andrew Warner talks with Henry Ndubuaku about Needle, Cactus Compute’s tiny on-device LLM built for reliable automation without requiring an internet connection or external API calls. Henry demonstrates how Needle turns natural-language requests into structured tool calls, powers software features and hardware like robot vacuums, and uses tightly scoped actions and confidence scores to reduce mistakes. They also compare Needle with Jev and explain why running an 8–29 MB model locally can matter for privacy, cost, and bringing AI into smaller devices.

Links featured:

Needle: https://github.com/cactus-compute/needle

00:00 - Needle: Run a small, free LLM locally for private and reliable automation.
00:18 - Smart Home Demo: Turn lights on and off by translating natural-language requests into structured tool calls.
01:39 - Search & Matching: Use Needle inside software for reminders, search, extraction, and other defined actions.
03:18 - Local App Integration: Add Needle to software without relying on Anthropic or another external LLM API.
04:12 - LLM.txt: Give Claude the context it needs to incorporate Needle directly into an app.
04:48 - On-Device AI: Put Needle into hardware instead of limiting it to desktop software.
05:06 - Robot Vacuum: Control a vacuum with natural-language commands while keeping the model on the device.
06:09 - Zapier SDK: Add integrations with thousands of apps alongside Needle-powered local automation.
07:12 - Reliability: Restrict Needle to carefully defined tools and actions to reduce unexpected behavior.
08:51 - Confidence Scores: Use Needle’s confidence output to decide whether an automated action should execute.
10:12 - Phones & Laptops: Organize photos, start timers, copy screen content, and automate device actions locally.
11:42 - Needle vs Jev: Compare their shared focus on constrained outputs and safer automation.
13:21 - 8–29 MB Models: Run Needle on small devices like smartwatches, remotes, AirPods, and other privacy-sensitive hardware.
14:15 - GitHub Origins: Andrew explains how Needle first caught his attention after trending among the week’s top GitHub repos.

👉 Media/Sponsorship Inquiries: https://thenextnewthing.ai/l/sponsor

👉 Join us: https://thenextnewthing.ai/