James Ulan 🗽

@Trace_Cohen Could never do that.

I don’t understand people who order food, eat half and don’t take the rest home.

Claude code is much better than Codex for non-technical vibe coders.

Think I just generated a slug of emissions equivalent to New Zealand's annual Co2 output by asking Perplexity Computer to update a PowerPoint slide for me.

Don't forget to say 'at scale' at the end. Sounds way better.

@shcallaway @aaron_epstein Very cool

Mom, I'm on the YC YouTube channel! It was an honor and a pleasure to sit down w/ @aaron_epstein to talk Sazabi, doing YC for a second time, and lessons learned from my first company Opkit (S21). Aaron has followed the full arc of my founder journey. In our conversation, he teases out some important insights and advice that will be useful to founders of all kinds. Thank you for having me! youtube.com/watch?v=UGYnC9…

The S&P 500 is now 9.2% percent off it's peak. What stocks are you buying?

@HarryStebbings 100% agree. What a terrible way to start a relationship and a bad signal.

The worst mistake people make when hiring someone new. They try to negotiate small amounts on salary. “The candidate wanted $100K but I got them down to $95K.” NO. BAD. Pay them what they want. They come in stoked. Ready and wanting to crush. Don’t be cheap on this.

@sohan_zhang Not cool. Have some human decency, investors!

one of the most frustrating experiences about the fundraising process is how decent respect gets thrown out the window one VC fund scheduled 3 meetings with myself and the team over 3 weeks, the last of which included an in person meetup that I took the courtesy of meeting them in the hotel they were staying at in SF. after a conversation, they mentioned that they would get back to me at the beginning of next week. cue 5 days later and i ask for an update. ghosted. i sent follow-up updates, as requested, over the span of a week, and still no response. i completely understand rejection, but it's very frustrating to take multiple meetings over weeks and get completely no response. future YCombinator founders, check the Bookface for investor reviews.

@shcallaway @sazabi Congrats on the launch!

Introducing: Sazabi (@sazabi). The AI-native observability platform for fast moving engineering teams. Backed by engineering leaders from the world's top AI and dev tool companies: Graphite, Vercel, Browserbase, LangChain, Browserbase, and more. Sazabi is taking a radically different approach to observability, centered on three core principles: 1. LESS IS MORE Product engineers don't need more observability bells and whistles. They need clear, actionable answers. The best UX for observability is chat. 2. LOGS ARE ALL YOU NEED The "Three Pillars of Observability" is an outdated idea. Logs can do everything traces can do, and more. 3. MONITORING IS DEAD Creating static monitors for fast-evolving systems is a fool's errand. The future is something different entirely: agentic anomaly detection. I started Sazabi out of desperation. After years of working on observability systems at Crunchbase, Brex, and 11x, I was all too familiar with the toil and complexity. Dashboards. Monitors. Session recordings. Error tracking. APM. RUM. Instrumentation. OpenTelemetry. Endless screens and configuration. It was too much. And despite everything, my teams still struggled to find and fix production issues. I was fed up. I dreamed about a better way. AI makes it possible. Today, I'm proud to introduce Sazabi to the world and announce our closed alpha program. We're selecting 10 startups to participate. Learn more: sazabi.com

@GJarrosson Great topic. They incinerated so much money too. WeWork and many others. But made incredible bets as well.

Before the Vision Fund, venture capital had a standard rhythm where a $50 million round was considered massive. SoftBank ignored those rules and started writing $500 million checks as a baseline. This massive influx of cash forced every other growth investor to adapt or die. If you weren’t willing to match SoftBank’s inflated valuations, you simply lost the deal. The result was a total recalibration of the late-stage market where burn rates exploded and profitability became an afterthought. We are still living with the consequences of that shift today.

@ajhodls partly a function of YC having them take meetings starting only the week before.. whereas past classes would take meetings 2 weeks before

most of the YC founders I talked to hadn't come close to their fundraise target by demo day. i suspect this can be extrapolated to the median company in the batch as well.

@guilleflorvs @MendralHQ @FenrockAI Yes. All 3 of them.

@JamesUlan100 @MendralHQ @FenrockAI interesting! thanks! Did you invest in any of them?

Which are your favourite W26 YC companies?

Nobody is talking about this but they should. The greatest short-term risk to AI’s progress is whether China invades Taiwan. When, not if, China moves on Taiwan, TSMC may shut down operations, which would slow or halt Nvdia chip deliveries. This would dramatically limit the amount of available compute at a time when society is using more and more every week. Model training would become more expensive. And people on $20/month plans will see their usage dramatically curtailed. Reliability could worsen for all users, including startups, degrading quality of their services. OpenAI, Anthropic and Gemini may raise prices which would increase burn for startups. The winners will be edge compute and startups that reduce token usage. Western, Korean and Japanese chip fabs will benefit too, as governments pressure them to build chip production capacity. Nvdia could become a loser if extreme scarcity drives investment in more efficient chips and data center networking architecture. Even worse, if China successfully takes over TSMC they could redirect some of TSMC’s capacity to their domestic champions.