If code has low value. Code from uncontrolled internet sources has negative value. I'd rather have my own cursor say stupid things than a random internet entity.

millennial gamers are the best prepared generation for agentic work, they've been training for 25 years

If you read this til the end, the chance that you’ll give up on entrepreneurship is 81% or you gonna learn the things that helped me keep going & get here I’ll touch on > product-vs-marketing debate > if distribution is the king > the truth I learned the hard way Let’s go If we ignore the enterprise sales, then it’s impossible to grow at a decent rate without word of mouth effect. WOM happens if users love the product. Making smth users gonna love is the hardest job on earth, same as making a song that tops the charts. People love if you solve the problem that they care enough about in a way that impresses them so much that they wanna brag about it to their network their whole status spins around them discovering & consuming high quality cool products, and if your product falls into this category then users share it to earn status. Once the facts above are in place, the product ends up dominating its niche. How to build such product in b2b: > the general idea must be very relevant for the users and it must >> make them more money >> save them serious amount of time >> remove serious amount of stress > while you build the product, there are 1000 new little ideas you need to come up with for little flows and use cases > just like in a great song, all of it must be great, not just a little part of it > in most cases you need to reiterate on existing features > basically remove your code and write a new one > which is painful and you start adding new features instead of reducing their number and reiterating with a few of them) The quality of your intuition is a dealbreaker here, because all those ideas have so many A/B/C decision to take Your intuition can be hacked by you asking those questions directly from your users Talking to users isn’t that easy, in most cases users give wrong answers To get right answers you rather need to talk to them about their problems and then magically come up with solutions Throw those solutions to your users and observe. If you see them complete the flow or come back to you saying “wow, great feature..” then you’re in the right path All of this takes months or years, and while you iterate the market may just change or move forward and your whole product is obsolete It means your innovation/iteration speed must be rapid, which is super difficult too Ofc u need an initial push, to get some users to test your product on, and in almost all cases this is done just like Airbnb did(stupidly go around knowing the doors, and being okay with such incapable marketing effort) So basically there is no “first I come up with an idea that I think is great, then I build the product and then I go market it” All 3 happen simultaneously in one single process that’s impossible to even separate, That’s why 99% of the startups are done by founders first, and they turn it into process-based system only after PMF, Before PMF they are just swimming inside the ideation/building/marketing maze as it’s one inseparable flow Long story short, building successful business is hard and you must learn to love the process, because the outcomes most likely gonna be meh for most of the time, if not ever. And if your product isn’t doing well, most likely it’s everything to blame simultaneously: the idea, the product, the features, the UX, the niche, the timing, the marketing… Improve everything. Don’t assume you’re good at any of this unless you’ve success in the past.

So apparently @GeoffreyHuntley's Ralph Wiggum technique (and our Claude Code plugin for it) went a little viral over the holidays? I thought I'd share a typical prompt I use with it. All of the instructional videos I've seen don't give thorough enough completion promises, IMHO

Also my experience. Most notably: i stopped using tailwindcss.