Arisay

@tigawanna @t3dotchat @zero__ms Everything work out of the box, you just save your API key in the in-app settings page and the server will encrypt it with AES-256-GCM so its always secure and never exposed

@ArisayAL @t3dotchat @zero__ms What's the best practice in storing keys in a BYOK set-up?

I’m open-sourcing Rift, a lightweight AI chat for teams. Think @t3dotchat , but self-hostable and built for organizations. • BYOK + Zero Data Retention for AI Providers • Super fast navigation powered by @zero__ms • File conversion to Markdown + RAG • Nested conversation branching • Org-level policies for model access, tools, and knowledge • Mobile app coming soon! And the backend is built on @EffectTS_ v4. We wanted to get the chat fundamentals right as there are still lots of users that prefer a clean chat UI for some tasks. Check out the repo, you can self-host it on @Railway in literally 5 minutes ↴

Totally fair, I actually used OpenWebUI for some time before starting Rift. For my use case, I wanted something lighter, faster, and under a more permissive OSS license, so Rift came out of that. You can try it for free, and I’d genuinely love to hear what feels worse or missing, just note that we won't have all of the OpenWebUI features rift.mx

@ArisayAL @t3dotchat @zero__ms if only we weren’t already on openwebui :)

They’re great! They give you ~$20 in usage, and they are my inspiration. But I’ve seen a lot of people who like T3 Chat and still don’t use it because their messages aren’t encrypted. Self-hosting Rift gives them confidence that their data is secure on their own server. We also offer a $8 plan with ~$8 usage for our hosted version! rift.mx

@ArisayAL @t3dotchat @zero__ms T3 chat is so cheap that it isn’t really worth self hosting. $8 for access to a ton of models with an outstanding image generation UX. That’s hard to beat. Self hosting is good but I want to get the latest model that comes out live in my instance without thinking of upgrading

@lassevjl @t3dotchat @zero__ms Weird, that shouldn't happen. Mind sending me more info to DM so I can fix it? The first thing that comes to mind is that there may be an org policy enabled in your self hosted instance that disables OpenAI models.

@ArisayAL @t3dotchat @zero__ms Why cant you use openai models on self hosted?

@itswillbrazil @t3dotchat @zero__ms It did! I think it’s great for a pilot. It explained the repo overview well, and it was actually kind of fun to listen to.

@ArisayAL @t3dotchat @zero__ms @ArisayAL - I’m getting code2cast off the ground and your feedback about the quality of the podcast would be super appreciated if you have 3 min to give it a listen. Did it do a good job explaining your codebase on the pilot episode?

@ojusave @t3dotchat @zero__ms @render Of course, that sounds great!

@ArisayAL @t3dotchat @zero__ms This is cool . Would you be open to adding a @render integration ? Happy to send you some credits via DM if you want to test

Heres the repo! github.com/Compound-inc/r…

Everyone is lying to you. I’m sorry. You’re building a product, validating an idea, and hearing endless “positive” feedback, yet no one actually wants to use it. Why? Because most people won’t tell you the truth directly. And most founders ask the wrong questions in the wrong way. That’s why The Mom Test by @robfitz is my favorite book to share with entrepreneurs, builders, and designers. Read it.

Apps with usage limits tied to subscriptions can be abused if usage resets at the start of a billing cycle. A user can: Subscribe → get 100% usage → use it all Block the card at renewal Subscription goes Past Due App resets usage anyway User keeps consuming during the grace period without paying Then once the subscription fully fails and is cancelled by the billing provider, they resubscribe and repeat. They’re basically getting free usage for the entire grace period.

@nizzyabi That's amazing Nizzy! Congrats to the whole team on the launch!

@grimcodes Port it to WebAssembly

yuhh we playin block game

Crazy that adding a nested branches feature for a chat requires studying Git

Been waiting for this small badge to appear on OpenAI models for so long. Thanks @vercel_dev

@0xDecker Looks great!

new landing

Got access to @orchid_hq a few weeks back. Wild how little I actually open my email client since. We've been calling everything an "AI assistant" for years, but let's be honest, most are just fancy autocomplete you have to babysit. You feed it context again and again, and it goes in one ear and out the other. You end up doing more work managing the tool than using it. Orchid is the first one that actually feels different. For weeks now, I just approve or correct my emails. I've even forgotten I used to tag them. It's been great. Now I have more time to ship instead of burning mental energy on drafting replies, answering emails, or creating Linear issues for bugs. The future is bright!

@chribjel unsubscribe

unsubscribe

@DalasReview Si estás usando react native te recomiendo usar generaltranslation.com, están en beta pero tiene muy buen DX y un CLI que traduce automáticamente a todos los idiomas bajo demanda

De las cosas más tediosas y pesadas (y que más he odiado) de programar tanto Eskiwi como SeaCret es traducir ambas Apps al inglés 💀 Recién ayer traduje Eskiwi y son 1200 entradas (algunas más largas que otras) y es un trabajo de revisar línea por línea y lo odié porque no es construír nada nuevo ni representa ningún reto para mí, ni aprendo nada. Simplemente es copiar lo que ya escribí, meterlo en un ID, pensar en ese ID nuevo, copiarlo en el translate, y corregir algún posible error de traducción. Y no, para los que preguntan, la IA NOOO puede hacerlo. Copilot me estropea el código y se inventa mamadas sin sentido cuando hablamos de una App que tiene como +200mil líneas y empieza a alucinar. Lo probé y sólo me estropea el código y se salta corchetes, se inventa llamadas, no respeta las constantes... etc. Y para los que dicen que yo no sé programar y me lo hace todo la IA: chúpala jodido imbécil. Eres un fracasado de mierda que en su vida ha programado un Hello World.

@raphaelsalaja Already read everything. So much to learn!

userinterface-wiki is completely open source. wondering how i made something, or just want to see behind the scenes, it's all available here. github.com/raphaelsalaja/…

@mannupaaji @vercel Hope you weren't using the Turbo performance Build Machine like I was. Fastest way to burn the $20 credits

Hey @vercel please don't bill me for this, bill GitHub instead