Garry Tan, the notable CEO of Y Combinator, recently captivated an audience at SXSW by revealing an intense drive, bordering on "cyber psychosis," fueled by his profound excitement for working with AI agents. He admitted to barely sleeping due to this passion.
During an on-stage interview on Saturday with fellow venture capitalist Bill Gurley, Tan stated, “I sleep like four hours a night right now.” He continued, “I have cyber psychosis, but I think like a third of the CEOs that I know have it as well,” a remark he made in jest about his current AI obsession. It's worth noting that AI-induced psychosis can, in reality, be a serious condition.
Tan described the transformative power of AI, likening it to recreating a past startup venture. “Once you try it, you’ll realize: it’s like I was able to recreate 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 — like I remember, you know, sort of being on like modafinil,” he recounted, referencing the wakefulness-promoting drug popular in startup hustle culture. Tan had previously sold his Y Combinator-backed blogging platform, Posterous, to Twitter in 2012.
Now, his engagement with AI agents has naturally amplified his mental state, leading to an organic form of 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 explained. “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.”
His enthusiasm for these AI agents culminated on March 12, just two days prior to the interview, when he openly shared his Claude Code (CC) setup on GitHub under an open-source license. This setup featured six "opinionated" Claude Code skills he had developed, which are essentially reusable prompts stored in specific "skill.md" files.
“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, naming his Claude Code configuration "gstack."
Gstack is publicly available, operating under an open-source, MIT license. Tan emphasized its ease of installation, requiring just one paste to set it up on a local Claude Code environment and a second for integration into a team repository.
Since its initial release, Tan has continued to expand gstack, adding numerous skills. While the gstack GitHub repository currently lists 13 skills, Tan's frequent updates on X suggest constant evolution.
He provided an illustrative example of gstack's functionality in one post: first, Claude acts as a CEO to evaluate a startup idea or feature; then, another skill directs Claude to write the feature as an engineer; finally, Claude reviews its own work for potential bugs and security vulnerabilities in the role of a code reviewer. Additional skills within gstack cover areas such as design and documentation.
The initial reception for gstack was overwhelmingly positive. Tan’s announcement tweet rapidly went viral on X and quickly became a trending topic on Product Hunt. The project quickly garnered nearly 20,000 stars on GitHub and saw 2,200 "forks," indicating widespread adoption and modification by developers.
However, shortly after gstack's release, a subsequent tweet from Tan sparked considerable controversy and backlash.
He shared a message from a CTO friend who lauded gstack as "god mode" for its ability to instantly uncover a security flaw in the friend's company code, predicting its widespread adoption.
The CTO friend’s text read: "Your gstack is crazy. This is like god mode. Your eng review discovered a subtle cross site scripting attack that I don't even think my team is aware of. I will make a bet that over 90% of new repos from today forward will use gstack."
This claim ignited a storm of criticism. One founder, among many, posted on X: “1) Garry should be embarrassed for tweeting this. (2) If it’s true, that CTO should be fired immediately.”
Vlogger Mo Bitar offered a critical perspective on gstack in a piece titled “AI is making CEOs delusional,” pointing out that the project primarily consisted of "a bunch of prompts" within a text file. Bitar articulated a common complaint among developers: those who already utilize Claude Code often have their own personalized versions of such setups.
Another commenter on Product Hunt directly challenged the project’s prominence, stating, “Garry, let’s be clear and honest: if you weren’t the CEO of YC, this wouldn’t be on PH.”
Given the divided opinions, the question arises: is gstack a uniquely valuable tool for working with Claude Code, or is it merely unremarkable? To gain clarity, experts were consulted, including Claude itself (which, predictably, offered an overwhelmingly positive assessment). ChatGPT and Gemini were also queried, both providing surprisingly favorable reviews.
ChatGPT's opinion described gstack as a collection of “reasonably sophisticated prompt workflows, but they’re not ‘magical.’” It added, “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 characterized the setup as “sophisticated,” further elaborating that “gstack is essentially a ‘Pro’ configuration. It is less about making coding easier and more about making it correct.”
Claude itself praised gstack as “a mature, opinionated system built by someone who actually uses it heavily,” concluding, “It’s one of the better examples of Claude Code skill design out there.”
These expert evaluations from the AI models themselves generally signal approval for gstack.
On Monday, Tan further articulated his passion 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.”
Despite multiple requests for comment, Tan did not provide a response.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.