@catalinmpit Uno reverse your thinking - incumbents means there's a market and willingness to pay (unless they are freebie apps!). Then research..market big enough for you to take a small slice too? anything to improve on incumbent offer? any sub niches? Learn, fail/succeed, repeat.
@petergyang I plug in @mobbin MCP with my Claude code and ask it to come up with shortlists of the best 'x', then make mockups aligned to our app's brand identity inspired by them. Early days but really effective so far, you get some good designs and labels where each part is inspired from.
@BuildWithxAI 100% agree on the speed change, Iโm still not sure the launch and distribution principles are different, just the margin for error is smaller. We can build faster now, but we still need the same go-to-market fundamentals to actually win.
@IMJustinBrooke When I excitedly speak to my friends about it (tech sales, medical marketing, physio, commercial real estate), the vast majority of them glaze over and say, "Yeah, it sounds cool, but the thing is, I'm not really interested in working with it. I just want the thing that works."
@levelsio I think you're missing all the stuff below the iceberg when you introduce a system, at least into a company, especially if it's an operating system or something that does more than complete one single rote function. Keep faith.
@Da7_Tech I struggle to understand the workflow when people are using a split between Codex and Claude. So you point out your repo and get it to work on different features within it, then ask the other to review?
@keean_edward Amen - I am building for small underserved, non technical niches in the UK and able to offer something tailored and feature complete that can be a third of the price of the enterprise leader. The boundaries of the software universe are ever expanding.
@claudeai Win! Now can we get the Fable 5 advisor fixed? It's returned "unavailable" every session since launch, over 2 weeks now. Repro's sat in #73365, still open. It's the one I hit every day.
Advisors in Claude still doesn't work and there's an open issue on it with others confirming the same. When will one of your agent loops resolve this?
@bcherny
@mattpocockuk My base is Opus 4.8 High. For what I'm working on right now I feel with the max usage I'm not hitting the limits. Interesting to see you are using medium. How big really is the difference between medium and high?
@abossy@claytoncorreia@lennysan I'm having this question myself, having painstakingly created my own task manager with the naรฏve expectation that it would be important but I rarely look at my task manager any more. I run it off my open Claude Code sessions and the session logs to pick up from the last point.
@paulg Very true, it extends to almost all modern knowledge work. E.g. consulting - client deliverables are rarely built from scratch. They are assembled from frameworks, templates, and playbooks. Is anything truly 'original'?