


software should feel good to use


what helped: - @iamharaldur telling me I need to think less and do more - setting up the feedback loop (for myself) of posting about product = products sell. ton of low-hanging fruit here bc I just haven’t posted as much as I need to - ai lowering the barrier on the parts I avoid due to emotional baggage (making IG videos) - just getting older and having more life experience. accepting my flops as necessary origin story and kinda fun in a masochistic way. might flop at this next thing but how can I make that time as fun and rich as possible. being stupid is good material for stories. detaching my sense of cause (perfect input) and effect (my goal) - remembering literally nobody cares. the longer i live the more i regret not going for certain things in the past. i know i still have so much more in the tank - seeing new entrants catch up fast bc of AI reminds me i gotta always be reaching for more there’s no time to stay frozen
unlocked a big mindset shift lately from "i have to do it perfectly because i want to do it right. but the bar is too high so i won't do anything at all" to "i want X outcome. how do i get people to pay attention so i can get X outcome"

T3 Code just hit 31.7 Trillion tokens in a day. Over $10,000,000 in API spend in a 24 hour window 💀



since some of you are apparently becoming smooth brained made a fork of T3 code with my own harness what’s the benefit of the harness? nothing.. it just forces you to solve puzzles before you get any response so your brain gets some wrinkles the worst captcha ever, try it 👇

I find myself giving this advice pretty often, figured I should share it publicly Once you're paid enough to live, the ONLY thing that matters is that you're surrounded with awesome fucking people.


I am slowly becoming known as the guy that gets excited about blimps, and I am so here for it. cc @Dan_The_Goodman



T3 Code 2.3 is out on iOS and Android 📱 🎙️ Talk to your agents • On-device dictation, now on Android as well • Or transcribe on your own computer through a paired environment • Or opt into cloud dictation with any model from OpenAI, Grok or OpenRouter 💬 In the app • @-mention a model to name it for the agent • /side starts a side chat without leaving the thread • Subagent approvals show up in the parent thread • Mermaid diagrams and math render like on the web • Long-press an image to save it ⚡ Faster sync and reconnects, even on bad Wi-Fi, and threads open faster 📐 New workspace layout system with better support for different poses across phones, foldables, tablets both on Android and iOS Get it today from the App Store or Google Play

I am just imagining how much more interesting robberies will be if they have to fight a bunch of drones
Meet Ring Drone Cam 🚀 Unlock a new way to take command of your home. Fly room to room for unobstructed aerial views. Use Live View to check in on different rooms or create custom patrol routes unique to your space, right from the Ring app. Learn more: to.ring.com/DroneCamX

Small new feature in the next @t3dotcodes nightly: Text Generators Thread titles, commit messages, PR descriptions and transcript cleanup no longer have to spin up a coding agent. Point them at any model API instead ¹: OpenAI, OpenRouter, Anthropic, or any OpenAI-compatible endpoint (Ollama, LM Studio…). Useful if you for example want Muse to cleanup your voice transcriptions but don't have a Muse Code sub and don't wanna setup a Pi or OpenCode provider just for some basic completions. ¹: Uses API prices and not subscription usage. Obviously you can continue to use the agent harnesses as before to use your sub quotas


You can now send off transcription to select cloud providers if you don't wanna run local models! BYOK for OpenAI, Grok and Openrouter will land in next nightly! Note that this does not work with your existing subscriptions!
You can now run local transcription models on your @t3dotcodes environments and use it as a dictation hub for any client! Choose among an abundance of models, filter through which best suits your needs, dictionary for words the models don't get right for your voice, and optional customizable post-processing! Try it out and send feedback!