Tech
Why Garry Tan’s Claude Code setup has gotten so much love, and hate
Y Combinator’s famed CEO Garry Tan told a SXSW audience that he’s got “cyber psychosis” and is barely sleeping because he’s so excited to be working with AI agents.
“I sleep, like, four hours a night right now,” he told his interviewer, fellow VC Bill Gurley, during an onstage interview Saturday. “I have cyber psychosis, but I think a third of the CEOs that I know have it as well,” he joked about his current AI obsession. (At least, we hope he was joking. AI-induced psychosis can actually be a dangerous condition.)
“Once you try it, you’ll realize: It’s like I was able to re-create my startup that took $10 million in VC capital and 10 people, and I worked on that for two years, and I took anti-narcoleptics — I remember, you know, sort of being on modafinil,” he described, referencing the sleep-preventing drug that’s popular with the startup hustle-culture crowd. (Tan sold his Y Combinator-backed blogging startup Posterous to Twitter back in 2012.)
But now, his psyche is so amped working with AI agents, it’s a natural insomnia.
“I don’t need modafinil with this revolution. Like, I’m up. I slept at 4 a.m. I woke up at 8 a.m.,” he said. “I wanted to sleep more, but I couldn’t because: Let’s see what’s going on with the 10 workers. I’ve got like three different projects going right now.”
He’s so excited about his agents that on March 12, just two days before the interview, he proudly, freely shared his Claude Code (CC) setup on GitHub under an open source license. The setup included six “opinionated” Claude Code skills he developed. Skills are reusable prompts stored in special “skill.md” files that instruct the AI how to behave in specific roles or tasks.
“I’ve been having such an amazing time with Claude Code, I wanted you to be able to have my *exact* skill setup,” he posted on X. He called his Claude Code setup “gstack.”
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
Since then he’s added several more skills. The gstack GitHub repository currently lists 13, but it seems like every hour Tan tweets about something new.
In one post, he gave an example of how his setup works. First, he gets Claude’s opinion on whether a startup idea or feature is a good idea using a skill where Claude acts like CEO. He uses another skill to have Claude write the feature as an engineer, and another to review its own work for bugs and security issues as a code reviewer. Other skills cover design, documentation, and so on.
The love for gstack began immediately: His tweet went viral on X and trended on Product Hunt. It’s accumulated nearly 20,000 stars on GitHub with 2,200 “forks,” meaning people who have taken the files to modify for themselves.
But shortly after releasing gstack, Tan posted a tweet that caused a heap of hate, too.
He wrote that a CTO friend told him gstack was “god mode” that instantly found a security flaw in his company’s code and predicted it will be widely used.
To quote just a few of the many hater comments that followed: One founder posted to X: “(1) Garry should be embarrassed for tweeting this. (2) If it’s true, that CTO should be fired immediately.”
Vlogger Mo Bitar did a take on gstack called “AI is making CEOs delusional” in which he pointed out that the project was essentially “a bunch of prompts” in a text file. The vlogger summarized the common complaint: Developers who use Claude Code already have their own versions of this.
Added one person on Product Hunt, “Garry, let’s be clear and honest: if you weren’t the CEO of YC, this wouldn’t be on PH.”
So who’s right? Is gstack a uniquely useful way to work with Claude Code? Or unremarkable? To find out, I asked the experts, including Claude (which, not surprisingly, absolutely loved it). I also queried ChatGPT and Gemini, both of which were surprisingly positive.
Gstack is a group of “reasonably sophisticated prompt workflows, but they’re not ‘magical,’” ChatGPT opined. “The real insight here is that AI coding works best when you simulate an engineering org structure. Not when you just ask: ‘build this feature.’”
Gemini called the setup “sophisticated,” adding that “gstack is essentially a ‘Pro’ configuration. It is less about making coding easier and more about making it correct.”
Claude called gstack “a mature, opinionated system built by someone who actually uses it heavily,” adding, “It’s one of the better examples of Claude Code skill design out there.”
We’ll take that as a thumbs-up from an expert on the topic.
On Monday, Tan explained in another X post, “I took modafinil just to stay awake longer to be able to turn the momentary crystalline structures I had in my brain into lines of code before sleep or human distraction turned it to grains of sand. I love coding but I love coding with AI even more. I speak it listens and we create. I see the structure and it is built. There is no more powerful an experience to me than that.”
Tan did not respond to multiple requests for comment.
Tech
Gamma adds AI image-generation tools in bid to take on Canva and Adobe
Gamma, a platform that lets you use AI to create presentations and websites, is launching a new image-generation product for making marketing assets as it seeks to better compete with the likes of Canva and Adobe.
The company says its new product, called Gamma Imagine, will let users employ text prompts to create brand-specific assets like interactive charts and visualizations, marketing collateral, social graphics, and infographics. Gamma currently provides more than 100 templates, which you can use alongside its AI tools to build the kind of assets that you need.
To power its data-driven asset-generation features, the company is integrating with tools like ChatGPT, Claude, Make, Zapier, Atlassian, n8n, and Superhuman Go.
“As we started working with a lot of our early users, we realized that in the presentations they want to create, there was a variety of graphical design use cases that they all also had,” Grant Lee, Gamma’s CEO and co-founder, told TechCrunch. “So we worked alongside them to develop basically a new set of tools that allows them to go far beyond just the traditional presentation format,” he said.
Lee believes Gamma sits well between tools for professionals like Adobe or Figma, and legacy tools like Microsoft PowerPoint.
“We think we can serve the very long tail of knowledge workers and business professionals whose demand for their job is to communicate visually, but they just don’t have the tools. They need to pull in a design resource to be able to help with this stuff, and we want to make an-AI native approach that serves their needs in the sort of middle that we feel is really underserved,” he said.
Last November, Gamma raised $68 million in a Series B round led by a16z, at a $2.1 billion valuation. At that time, the company said it had ARR of $100 million and 70 million users. The company told TechCrunch that it is approaching 100 million users now.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
Tech
Google’s data center power playbook comes into focus
Google may have signed on to President Trump’s toothless power pledge, but it’s clear the company started working months ago on a framework to power its data centers.
On Thursday, Google said it will work with Michigan utility DTE to add 2.7 gigawatts of “new resources” in suburban Detroit to power a new data center in the region. Some specifics are still fuzzy at this point, but the deal mimics one signed last month with Xcel Energy to build a data center in Minnesota. This is how Google will develop new capacity for its future data centers.
The new plan includes 1.6 gigawatts of solar power, 400 megawatts of four-hour energy storage, 50 megawatts of long-duration energy storage, and 300 megawatts of “additional clean resources,” which is a squishy way of saying anything from wind and hydro to nuclear and geothermal.
TechCrunch sent Google’s PR people a number of questions, and while they responded with some details, it’s clear there’s a lot to the proposal that either isn’t fleshed out or isn’t fully public yet. To wit: Does “clean resources” include natural gas? We haven’t received a reply on that one yet.
The remaining 350 megawatts of the 2.7 GW deal will be covered by demand response, which is when large electricity users curtail their use for brief periods of time. What shape that takes remains to be seen. Google may be looking for companies that are willing to dial back their electricity needs at certain times, or it will turn off its own data centers when the grid is strained.
The DTE deal will also use Google’s Clean Transition Tariff, which it has been refining over the past year or so. The tariff was previously used in Google’s deal with Xcel Energy. It’s intended to allow Google to pay a premium to specify the types of power it wants deployed while also encouraging utilities to incorporate such technologies into their long-range planning. Previous instruments like power purchase agreements were often treated by utilities as one-offs.
Google also said it is introducing a $10 million Energy Impact Fund intended to reduce utility bills, including by insulating homes. It sounds a lot like energy efficiency programs run by utilities, just with Google’s name on it. Whether $10 million is enough to assuage regular people’s concerns about rising electricity prices remains to be seen.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
This is the second “bring your own power” package that Google has touted, though it’s unlikely to be its last. In many ways, it’s not that different from the way the company has operated in the past. Sure, the tariff is relatively new, but Google has been investing in or developing new generating capacity ever since it vowed seven years ago to use 100% carbon free power.
The difference is that those projects tended to be announced on their own timelines. Now, we’re seeing the inverse — power projects that are in the works get announced along with the new data center. Smart marketing or something more? We’ll know in a few years.
Tech
BuzzFeed debuts AI slop apps in bid for new revenue
BuzzFeed, the U.S.-based media company known best for its quizzes, listicles, and, for a time, a Pulitzer Prize-winning journalism division, is reinventing itself for the AI era. At least, that’s the pitch.
At the SXSW conference in Austin, BuzzFeed co-founder and CEO Jonah Peretti introduced the company’s next media foray: a spin-off called Branch Office, which will explore artificial intelligence in consumer-facing apps designed for creativity and connection.
The new company is an extension of the experiments BuzzFeed has run for years using AI technology, Peretti explained, in a halting presentation that began with slideshow glitches, before moving on to app demos met with silence or a polite tittering.
“We’ve been working on this secretly for over a year, and we’ve learned a lot from the BuzzFeed platform about what is coming with new kinds of AI formats,” Peretti said. “Using AI is the way of connecting people, building community around these pillars of culture, and taste, and community.”
Bill Shouldis, a director of product at BuzzFeed and the founder of Branch Office, presented two of the company’s new apps: BF Island and Conjure.
The first product, BF Island, is a group chat platform offering features for changing and editing photos using AI. This is not exactly groundbreaking tech in and of itself, but that’s not the point.

The key feature here is not the AI toolset but the in-app library of online trends and memes, created by an editorial team, which could inspire users to create AI photos referencing blink-and-you-miss-it trends like the McDonald’s CEO taste-testing a burger or the “frame-mogging” drama. (If you don’t know what these are, you’re probably not the “very online” audience that’s being targeted.)

The other app, Conjure, is similar to BeReal — the once-a-day temporary photo app — except that it instead appears to guide users to take daily photos of things besides themselves. (As a reminder, BeReal didn’t stick, ultimately exiting to Voodoo after losing traction.) In the demo, for instance, the photo prompt was “What lies between the trees and the moon?,” leading the users to snap a photo of the night sky. A series of spooky images flashed on the screen, followed by a whispered, “What will you conjure?”

We don’t get it, and clearly the audience didn’t either. After the demo, a lone cough could be heard among the silence, followed by uncomfortable laughter.
Shouldis then noted that AI is involved in Conjure, too, as the app has an “AI spirit for a CEO.” (Again, what?)
Peretti also introduced Quiz Party, a social app that lets you take BuzzFeed quizzes with friends and share your results.
BuzzFeed’s underwhelming presentation comes only days after the media company shared that it has “substantial doubt” about its ability to continue as a business and was engaging in strategic conversations focused on fixing its liquidity challenges. The company, which had a net loss of $57.3 million last year, said it would focus this year on its Studio IP and new AI apps, like these.
But even the tech-forward audience at SXSW was not convinced.
As one person pointed out during the Q&A session after the presentation, BeReal had struggled to get people to come back after the novelty wore off. What would an app like Conjure do to combat the same sort of retention problem?
Shouldis said that the app would evolve “and have different types of things happening and not just be exactly what it is today.” He referenced the potential to integrate things like video, audio, and prototyping with Claude Code to build community.
The premise behind the new apps is not unreasonable: AI can lead to faster software development, which makes it possible for companies to more quickly iterate and keep people engaged.
“In a way, software is the new content,” Peretti noted.
Of course, before you can iterate, you have to attract users. With its new apps, BuzzFeed seems to have thought more about what AI can do than what people want to do with AI, which is not a recipe for success.
