i went from 30+ grok bots to 7.
at first i made a new bot for every little thing. after a while i couldn't remember which bot did what, and i kept asking the wrong one.
so i rebuilt it like a small company:
- one boss per category: shopping, finance, health, x, and a few more
- each boss has its own skills for its own area
- i only talk to the bosses
when a one-off job comes up, like a damaged order that needs a refund, the shopping boss spawns a temp bot and hands it the skills for that job.
temp bot rules:
- always grey, so i can spot it instantly
- the name says who spawned it and why
- when the job's done, i delete it
now i know exactly who handles what, splitting work is easy, and nothing piles up.
one bot for everything sucks. this is way cleaner.
what i still want from @bot:
1. spaces. a dedicated tab in the app where my bot builds a live, interactive page, a whole second interface to talk to it besides chat. like a calorie tracker where i snap a food photo right there and the bot logs it, no typing or voice notes. or a dashboard that refreshes on its own and shows my emails and slack visually, instead of a scheduled message in chat
2. a temp bot type. a boss bot can spawn it and delete it when the job's done, so cleanup isn't on me
3. calls on earpiece, not speaker by default
4. a voice typing option. when i talk to my bot via x, it sent as voice message when i just wanted voice typing. i'd love a mode where my voice only becomes text
t3 code remote control is honestly the best remote experience i've had
running agents on my laptop, steering everything from my phone
and it can even stream videos now. that's kinda crazy
beautiful experience, no notes
@theo@t3dotcodes
@agenticjay bro it's far more creative than that. Yeah, it can make insane 3D, but give it an open-ended prompt like this one. see this you will understand https://t.co/UsKHi5pDBb
@jal_eelll They actually do have the feature built into the app, they just haven't enabled it on their end yet. You can still ask your agent to turn it on for you, and once it does, you'll have full control.
@dabit3 bro the last contest of devin was unfair only ios got it anyway but yeah offers are good very very bad ask like this anyway i am asking could you give max promo pls yeah i am sorry to ask you like that
Gemini Neon is DOGSHIT. π Even GPT Luna beat it.
And this exposes a HUGE problem with Gemini models.
For a simple test, I made a Rust algorithm and asked both AIs to make it 40% faster.
π’ GPT: "40% isn't guaranteed until we benchmark it." Actually tested performance.
π΄ Neon: Confidently claimed 2β3x speedups WITHOUT measuring anything.
No latency tests. No proof. Just made-up performance estimates presented as FACTS. π€‘
This is my biggest problem with Gemini models: they confidently turn GUESSES into FACTS.
A model that admits uncertainty is infinitely more useful than one that confidently bullshits you.
Google, confidence β competence. π
And that's exactly why I can't trust gemini with coding tasks , how ironic, both claude and gpt models doesn't suffer from this problem, FIX YOUR GODDAMN AI MODELS GOOGLE
ITS REALLY BAD
Trust me this is really bad cot. @GoogleDeepMind@GeminiApp@googledevs
can somebody please stop these people
one rust prompt and suddenly neon is "dogshit"
and this got 59k views. what are we doing
yes i'm jealous. no he's still wrong
https://t.co/piFpXDgUfK
@antigravity@OfficialLoganK Can you please remove the separate weekly quota for Argon and add it to Neon, or combine the quota across all Gemini models?