As of today, I feel like two Claude Max 5x subs are much better value than a $200 ChatGPT Pro sub. Fable is the absolute GOAT for front-end design and Opus feels unlimited.
To really lock in the UI/UX for https://t.co/Devk25Hg1H i used this tweaks trick...
have your agent make a dev-only tweaks section that lets you preview any component with an editor to modify variables
say "add a tweaks section with a live preview of X component. wire up design tokens to sliders so I can change them and see what they look like. add a way for me to reset them and copy modified values"
then later on say something like "the tails on the bubbles dont look right, add all design tokens to tweaks"
and it one shots a little internal editor
not only was ChatMRR entirely built with agents, i made it pretty much exclusively from my phone with https://t.co/PrkwahOkgg no way i could have done it without this workflow
(idea not my own, i think fb made a whole library for this. agents just make it easy)
i'm building a new app and i'm at a point where i feel like we (my agent and i) need to have a proper design and component system because anytime i ask something, it just creates things by itself, doesn't bother to look at my existing UI patterns, including layouts, colors, spacings, etc etc, and just "makes up" stuff on its own, which creates lots of design drift as time goes by. and i kind of feel like it's costing me more time, more brain cells, and more tokens whenever i need to correct it
so i need to design a self-regulating mechanism where the AI will go look for the right UI patterns whenever i ask it to implement a feature, a little bit like as if we had our own storybook (i'm not using it unfortunately because i think we can have our own? if there's a strong use case or very good reason i should spend a little bit of $$ for using it, i'll consider. until then i want to have my own little internal knowledge base for components and UI patterns that my AI can consult)
and for the record i'm currently out of Fable and Astra usage so i set myself up with a minimal pi coding harness and i chose deepseek v4.1 flash as the main model. i think this type of mistake happens less when i'm using a powerful model, but nevertheless, i'm just thinking that the drift can still happen, hence reinforces my belief that i need this type of system
what does everyone think? do you relate??
i feel like it's the best use case that i'd definitely go for, but i still don't see a reliable way to do it?
many people keep talking about how there's a risk of breaking cache and, overall, things might end up more expensive/costly in terms of tokens
im pretty optimistic that this will get solved soon enough though
to the builder who's doing this: thank you for your service lol
Jev is an incredible model router.
Ask Jev which models can solve a user’s task, and route it to the cheapest one.
We have a new frontier model overnight.
thats really cool
ever since i started indie hacking, i went to a few indie hacker meetups offline and joined some of their online group chats too
i started out building web apps, then shifted to mobile last year, and over time i realized i just wasn’t looking for the same things as most people in those groups
for me, there isn’t much value in spending hours every week at a meetup unless the people there are your customers, potential partners, or people you can learn something very specific from
otherwise it’s very easy to spend an afternoon chatting, have lunch, talk about tools and tech, then go home having barely moved your own work forward
and i noticed a lot of the conversation tends to revolve around building
latest frameworks, new stacks, technical choices, what everyone is experimenting with
which is fun, but once you’re trying to make money from products, you start caring a lot more about distribution, positioning, marketing and sales
that was probably the biggest disconnect for me
i kept going for a while, then eventually stopped because i realized i was getting more value from talking directly to users, working on distribution, and just spending more time building my own thing
i do love the idea of being in the same room as people who are at a similar revenue stage though, especially other mobile app founders
i feel like we’d probably have a lot more in common, from organic acquisition, app store stuff, paid ads, and all the weird little problems that come with trying to grow an app