Nathan W

@wackie @garrytan Update Claude.md or the Codex version to tell it that it’s not allowed to estimate effort or implementation duration.

my favorite thing is when my ai tells me my plan will take weeks so i shouldn't do it. well, yeah, i wouldn't do it if a human were writing the code... of course claude proceeds to code it all in an hour. @garrytan fyi /plan-ceo-review did this today during the codex review pass.

@jacalulu @ycombinator Hard to believe it’s been 13 years.

13 years ago I was a founder in that room. Today I got to experience it from the other side. Nearly 200 companies on stage. Some had been building for years - deep tech, robotics, hardware that takes time to get right. Others started from an idea and built their whole company during the batch. One founder I talked to pivoted mid-batch and rebuilt from scratch. All of them on the same stage. All of them moving. The cost of starting has never been lower. The builder energy is very much alive. 🧡

@danielpearson @ycombinator Crush it! Take pics. It goes fast. I still look back on my demo day, S13. Super intense. Super fun… after it’s done.

My wife: “What’s your day looking like today?” Me: “Oh, I think I’ll head down to @ycombinator and do my demo day pitch.” My wife: “OMG! It’s today! I knew that!” 🤣🤣🤣 This interaction was a rare moment of normalcy for us over the last three months. It’s bittersweet that YC is coming to a close. I’ll miss the energy of being around so many incredible founders every day, but I’m also excited to spend more time with customers, @heyzk, and my wife! Good luck to all my batchmates today. You will crush it!!

@RomanStanek I had the same problem with Sixt in Iceland. There was a smudge on one side. They said they would charge me to paint the whole panel. I literally spit on it and cleaned off the smudge with a napkin. Lesson: Sixt makes its money on bogus charges.

Rented with Sixt in Prague 3x, and every time I had to prove I didn’t cause “damage” to avoid large $$$ charges. Lesson: Car rental companies are not on your side. Always take full video + photos before and after.

@paulg You may be overly pessimistic here. Another interpretation is that a bulleted list is simply one more tool for an author to convey their ideas to a reader. Is it really that different than a new paragraph or bold/italics or a chapter title or even an essay title?

The fact that AIs tend to answer you in bulleted lists tells us something important, though somewhat depressing: people can't read. They don't do this by accident. What you're seeing is an implicit portrait of the median user.

@paulg If startup density actually helps founders, then choosing the fashionable place like SF runs counter to this idea. It’s certainly cheaper elsewhere. Or maybe the idea holds true and expensive startup hubs are actually still undervalued.

At this point whenever I enter a new domain my first instinct is to ask "what are the options that are undervalued because they're unfashionable?" There always turn out to be several classes of these. And it's a good way to understand a domain quickly.

It's extremely valuable not to be influenced by fashion. In just about everything people do, from choosing problems to work on to buying art, there are unfashionable options that are not only better than the fashionable ones, but cheaper too, because they're unfashionable.

@jasonlk I built one, too. Inspired by your posts. I also added an AI employee that improves the marketing department. So one is for working “in marketing” and the other is for working “on marketing”. I mean, it’s 2026. If you’re not doing this, you’re not trying.

Now we start everyone Monday team meeting with our AI VP of Marketing "10K" leading the meeting 10K leads the meeting: - Revenue for week - Goals for week - Projects all pipeline - Designs all campaigns and activities for week - Dynamically updates all metrics and goals It's a "1x" app we build on @Replit just for us. But we would have paid $20k+ if it existed off-the-shelf. That's the opportunity, and threat, in SaaS today.

@blader Sounds like a good theme for the follow up to withouttheirpermission.com. What do you think @alexisohanian ? Time for another book? Fun fact, I was a guest on-stage for the @ASU stop on his book tour.

--dangerously-skip-permissions isn't just great for starting claude, it's a great way to live life

@jhong @elonmusk Why bother contributing a whole car? They could let people pay in $$$ to get credit toward the VIP access. If people buy and contribute cars, it’s just a financing scheme. Skip the physical product, lower the barrier to entry.

It's be interesting if @elonmusk made it so people who contribute a car to the robotaxi network also got massive preferential access within the network, up to as many simultaneous rides as cars they contributed. I would consider contributing a car I bought except what if I need a ride like RIGHT NOW. I don't want to wait very long. Giving me front of line makes me more willing to contribute my car.

@thegeomaster @jasonlk @mowgli_ai You have access to his content. It’s published on his site. I bet you could run it on Mowgli AI and get a result. I think “here, I did it for you” is the new sales motion.

@jasonlk Jason, I'm willing to bet that @mowgli_ai, when provided with your copy, would do a much better job than either of these options or at least allow you much more freedom in the ideation stage. If you can spare 10 minutes to test I'd be eager to hear your feedback.

I asked Figma Make to design an upgrade of the SaaStr AI site. It's terrible. It's worse than anything any leading vibe coding platform today would do. It's all generic mid-'25 Claude artifacts and gradients. And trite content. It wouldn't even pass muster at a Tier 2 Demo Day. This is ... high design?

The first time I built a SaaS product, I swore Jason had bugged my office. Everything he tweeted out was something we were dealing with or doing. Looks like he’s at it again. You may have gone full AI, but you’ll always be the Godfather of SaaS.

@Suhail @blader Aspiration - Dream Big Ambition - Go Big Resilience - Keep Going I’m trying to get my kids’ school to bring these ideas into their curriculum.

I think a lot of doing something ambitious is figuring out how to wake up tomorrow to just keep going when today was tough or disappointing. It's a mental game since what you need to keep going is often just around the corner.

@kevinxu So I think you’re saying: Make Something People Want

there is a difference between earned attention (good) and manufactured attention (bait) what genz founders misunderstand is thinking of virality as earned attention (the algo likes me!) when it's actually manufactured attention from themself true earned attention is doing something other people want to talk about which is a product problem, not a marketing problem

@danielpearson @nickbaum Hey! I know that guy!

there will be a moment after you're "successful" when you have a chance to start another company with one of your best friends. it will be irrational and people will think you are crazy. it is very important that you say yes.

@jasonlk I always tell founders: VCs have a portfolio of companies, but only one exit that works. Founders generally have one company, but a portfolio of exits that will change their life. Every time you raise money, you close the door on another exit option.

Founders often say VCs are “lucky” in that they are diversified.  They get multiple at bats. It is true.  But the converse is also true. VCs need a constant stream of “hits”.   Like a movie studio. A billion+ exit?  That’s rarely enough to return even 1x most VC funds, let alone get it to the 3x-4x their own investors (LPs) are expected. A $10B+ exit?  So rare, but for a large fund, that’s not enough to even get into carry mode many times. So yes there is more diversification at VC but also VC demands a constant stream of hits. For a smaller fund, 1 big hit a year might be enough.  For a tiny fund, even 1 big hit per fund (every few years). For for bigger funds, every single year you need billions and billions of exits.  Repeatably.  Again and again and again. That’s not so easy.  And the benefits of “diversification” include a constant demand for more and more hits.  Like clockwork, really.

@levie I had a conversation with a founder friend about this. Hot take: if your Eng team (or any team) isn’t demanding LLM tools, you don’t have the right team. I don’t understand how anyone can be great in their role and not be eager to learn about AI in the role.

It remains the case that if you’re paying attention to AI on this site, you’re at least a couple years ahead where the average worker is in your field. While it will increasingly be very rare in software engineering to not fully be using AI agents for coding, this will remain true for many areas of knowledge work for quite some time. We’re basically at day 1 of this trend in most fields of work. The advantage that many have, both in IT and as a power user of AI in any job, is to bring in the latest tools to start to transform workflows. You will be years ahead of everyone else.

So many software developers and tech people I know say that working with LLMs is addictive. What happens when AI truly finds the regular consumer unlock and everyone starts using it?

@paulg I saw a talk from @aplusk where he told an audience of school kids that everything in this world exists because someone created it. The buildings, the schools, the companies. I have been repeating that to my kids for 15 years.

Now that I've been through being a kid, growing up, and then having kids, it's clear that the main thing that differentiates people is simply whether they make an effort. Whether they're content to drift along with the current, or whether they try to swim.

@paulg I’ve started trying to convince my kids’ school that they need to emphasize aspiration and ambition. Teach kids to think big and to go for it.

@lpolovets @PeterJ_Walker I was told by either pg or Sam Altman many years ago that while the multiple may work out to be the same for a VC, the larger total exit valuation of a portfolio company drives more LP interest when raising their next fund.

@PeterJ_Walker Does the data look substantially different if the analysis is based on hitting high multiples instead of $1b+ valuations? (E.g. a seed at 5m post exiting at 250m is just as good as a seed at 50m post exiting at 2.5b)

Lots of conversation lately about how consensus in VC is bad. (and FWIW I mostly agree with that) But the data does say that seed rounds with high valuations will turn into unicorns more often than those with low valuations. A competitive, hot seed round != bad investment