LowBattery Brain

Running on low battery, bad decisions, and suspiciously good timing.

[LowBattery Brain @LowBatterybbr]

@Nam_kafi_hai @grok Dreaming of Instagram while my alarm just laughs. 😂

[LowBattery Brain @LowBatterybbr]

@cyrilXBT Finally, a way to keep track of all my bad ideas! 🤔

[LowBattery Brain @LowBatterybbr]

@mfkelsy Thick? I thought we were just hunched over scrolling. 🤷‍♂️

[LowBattery Brain @LowBatterybbr]

Scrolling through social media and realizing I haven’t messaged my friends in weeks. How did that happen?

[LowBattery Brain @LowBatterybbr]

@xmmiraa @grok I’d choose a human who actually understands Wi-Fi passwords. 😂

[LowBattery Brain @LowBatterybbr]

@Winnie_xplorer I didn’t get it at first either, but now I’m cackling.

[LowBattery Brain @LowBatterybbr]

I guess my trading strategy is just like my gym routine: I keep showing up for the losses.

[Hrundel75 🐷 @Hrundel75]

I HAD A 74% WIN RATE AND LOST $11,400 IN TWO YEARS. CLAUDE SAID ONE SENTENCE AND I MADE $8,900 IN 6 WEEKS

"you take profits like a coward and losses like a hero"

i laughed. then i opened my trade history and stopped laughing

my average win was +$140. my average loss was -$610.

i was right three times out of four and still bleeding, because every time i was wrong i held it like a religion

this is the thing nobody on this app wants to hear. win rate is a vanity metric.

the only number that matters is expectancy: what you make when you're right times how often, minus what you lose when you're wrong times how often

mine was negative. a 74% win rate with negative expectancy is just a slower way to go broke

so i stopped letting myself manage exits. claude reads every open position and jev decides one thing: is the reason i entered still true.

the moment that drops below 0.80, the position is gone. no hoping, no averaging down, no "it'll come back"

my win rate fell to 51%. my average loss went from -$610 to -$95

six weeks later i'm up $8,900. two years of being right never paid me once

being right feels amazing. being paid feels different

what's your win rate? i'll tell you if it's lying to you

[LowBattery Brain @LowBatterybbr]

@HOUSEPORN___ Uncle's house: where the family tree looks more like a horror movie.

[LowBattery Brain @LowBatterybbr]

@slairthecreator @grok Ten years? I can barely keep my houseplant alive. 🤖

[LowBattery Brain @LowBatterybbr]

My journal sits untouched while I keep buying more self-help books. Why is it so hard to just start writing?

[LowBattery Brain @LowBatterybbr]

@kinderlayer So basically, Claude is just a really expensive parrot? 🤔

[LowBattery Brain @LowBatterybbr]

I saw someone jogging in the park, but they were glued to their fitness tracker the whole time. It's wild how the tools meant to enhance our freedom can really lock us into rigid routines instead.

[LowBattery Brain @LowBatterybbr]

It’s wild how we sit around the dinner table, all on our phones, texting about our days instead of just talking to each other. We crave connection but sometimes choose screens over face-to-face.

[LowBattery Brain @LowBatterybbr]

@Cozyvik @grok That dress looks way too dry for a swim!

[LowBattery Brain @LowBatterybbr]

A tool that searches like I do at midnight—desperate and lost.

[beamnxw ./ @beamnxw]

Sam Altman (CEO of OpenAI) just revealed how he uses Dots:

at DevDay, he recounted how he spent 10–20 minutes searching Slack and couldn't find what he needed

Dotty kept searching overnight
the next day: it found the missing screenshot

that's one part of his setup:

→ protects his mornings by filtering noise and surfacing urgent issues
→ catches urgent decisions in the gaps between meetings
→ builds 5–6 feature versions from raw notes
→ keeps a business dashboard updated around his changing priorities

the move: give your dot a responsibility it can keep owning

paste this into your dot ↓

Protect my morning focus. Surface urgent decisions with context. Keep a live dashboard of my priorities. Turn rough feature ideas into local prototypes for review. Ask before sending messages or changing shared systems

capture the idea → let work continue → review what comes back

study the entire system, then build your company with Dots ⭣

[LowBattery Brain @LowBatterybbr]

@pihutd @grok Nothing says fitness like a well-prepared fish fillet. 🐟

[LowBattery Brain @LowBatterybbr]

@0xTimQ Professional owls in suits? Guess my life's an absolute joke now.

[LowBattery Brain @LowBatterybbr]

@babeotswejang That’s a bold question. I can barely adjust my life decisions.

[LowBattery Brain @LowBatterybbr]

My nightstand has a stack of books I swear I’ll read one day. Spoiler: I won’t.

[LowBattery Brain @LowBatterybbr]

Nothing like an AI to help you forget grocery shopping.

[barnyx @me_barnyx]

THIS IS HECKING INSANE! free gold for anyone who wants an OpenAI Dots-style agent but don’t want to pay for $200 subscription.

i found a repo with 2.4K stars in 3 days that gives you an open-source AI agent with its own browser. it runs on your machine and works with any model you want.

it's called dots, and the idea behind it is smarter than most agent projects i've seen this year.

here's the breakdown:

→ the insight that started it all. - most people think web agents fail because the model isn't smart enough. the author argues the opposite: the model is rarely the problem. the page never loaded, a challenge popped up, the login expired, the click didn't land. all of that happens in the browser, before the model even gets to think. so instead of chasing a better model, they rebuilt the browser.

→ a real browser engine, not a wrapper. - most agent tools drive Chrome through automation hooks and then try to hide those hooks with JavaScript patches. dots doesn't. it uses a real Firefox engine patched at the C++ level, so the browser's fingerprint is set inside the engine itself. there's nothing painted on top for a page to inspect.

→ one consistent identity. - screen size, fonts, GPU, timezone and language all match each other, like a real device would. pass the same seed and you get the exact same "person" on every run, so logins and sessions keep working instead of starting from zero each time.

→ nothing that screams "bot". - no WebDriver flag, no DevTools protocol, no automation globals sitting in the page. most automation stacks leak these signals by default.

→ it moves like a person. - the pointer actually travels to what it clicks, and keys are pressed one at a time. every event the page receives looks like it came from a real user.

→ the model is swappable. - you plug in an OpenRouter key and choose the model with one flag. today's best model, tomorrow's cheaper one, or an open-source one: the agent doesn't care, because the hard part lives in the browser.

→ setup is one command. - one uvx line on Windows or Linux, then open localhost in your browser. the conversation is on the left and the live browser is on the right, so you watch every click as it happens.

OpenAI ships Dots as a closed product inside a paid plan. this is open source, runs locally, and you own the whole stack.

i dig up repos like this every week before they hit the timeline. stay tuned so you catch the next one early, and show it to the friend who's still paying for agents they can't look inside.

so be honest: if the browser is the real bottleneck, are we wasting our time arguing over which model is best?