making rahn.chat, my first public release

LOG 202501140120 #log

while catching up on ai news on saturday morning, i came across the release of another new ai chat wrapper. it's pretty cool, i like the methodology behind it but the marketing on x/twitter is funny. it feels hyperbolic for a famous software engineer to go on and on about such a simple app. venture backed startup lab with real software engineers like bro i'm pretty sure i can build something like this on my 6 year old macbook using cursor.sh.. and thus rahn.chat was born.

i see a lot of smart and impressive people on x talk about how ai isn't going to actually take jobs and.. man... i just don't see it that way. so here i am using agentic ai to build something that seemingly took a small team of real engineers much longer.

using openai o1 for planning, cursor.sh and claude sonnet for development, i deployed in replit about 36 hours later.

don't get me wrong, my app isn't as good.. at least not yet. it only costs me money right now lol and i'm sure they're making racks. but i'm just some guy using ai, and despite not being an engineer, it feels like i have teams of them at my disposal now. i don't think the world is ready for what these ai models will be doing in just a few more months.. fun trying to keep pace with it all.

i'm pretty happy with what i've shipped so far with rahn.chat and plan to keep grinding away on it after hours and on the weekends, we'll see where it ends up. definitely my finest day-and-a-half build yet lmao.

i have an openai grant until end of February so i can offer free 4o-mini access which is cool. i'm planning to tackle payment integration and get some billing model in place so that i can continue offering some level of free access. i want to fix all the bugs first and continue scratching off things on my wish list.

you get stuck talking to my dog, rahn, unless you sign in, which also removes the token limitations on 4o-mini. from there you can input your own api keys (🍔>🔑) to access all of the models (🧠) and create your own instructions and characters (📓), all of which is stored locally on your computer. you can view all your previous chats using the history feature (🍔>🗒️).

the server doesn't keep any user data, all of the custom instructions, ai conversations, api keys, ect, are all cleverly and securely stored on the client device.

i have a long way to go still, but this has been a really fun build so far.

i'm really happy with how well it performs, despite some lurking bugs..

i can't do much better than that, lol.

okay maybe i can do a little better than that, but hey... the app that inspired me doesn't even support mobile and we've got PWA support 😉

stay tuned for more updates, i'm excited to keep building this out