lfpmb

everything is a skill issue

[lfpmb @lfpmb1]

Four months ago, I released Rewind on the App Store.

The idea was simple: what if you could capture something after it happened? Rewind keeps a rolling buffer of recent audio, so when someone says something important, you can tap a button, save it, and transcribe it.

Today, @Apple announced a feature for Apple Watch built around essentially the same idea.

And they called it Live Rewind.

Same concept. Same “Rewind” name. Four months later.

As an independent developer, that’s a pretty surreal thing to see.

You spend months building something because you believe it should exist, ship it to the App Store, and then a few months later see that same idea become a native feature at Apple, under a similar name.

Either way, it’s pretty cool to see something I believed was worth building become a feature at Apple four months later.

https://t.co/J4E9jvcOJz

[lfpmb @lfpmb1]

I have been testing agent prompts that encourage strong code quality while remaining clear, simple, and not overly restrictive; this version has worked well for some time. It helps the agent move through and examine the codebase more easily.

“Write code that is readable, maintainable, and easy to reason about.

Avoid excessive indirection: do not split logic into too many tiny functions if it makes the reader jump around unnecessarily. At the same time, avoid excessive inlining: move secondary implementation details into clearly named helper functions when they distract from the main flow.

Keep the primary logic path obvious. A reader should be able to understand what the code is doing without holding too much context in their head.

Declare variables at the smallest reasonable scope and as close as possible to where they are used. Avoid long-lived mutable state unless necessary.

Use encapsulation deliberately. Keep internal variables, helper functions, and implementation details private or local unless they are truly part of the public interface.

When refactoring, preserve existing behavior while reducing unnecessary complexity, improving naming, limiting scope, and making the code easier to follow.”

[lfpmb @lfpmb1]

@chetaslua Are tool calls what makes up the most of this time?

[lfpmb @lfpmb1]

Excited to share that Bruno Fonseca and I won the Flutter SF GenUI Hackathon with BayHop!

BayHop is a Bay Area transit and exploration app built with @FlutterDev and GenUI. Instead of returning plain text, the assistant streams real Flutter UI: trip cards, live departure boards, service alerts, route overlays, place recommendations, and itinerary planning.

The experience is powered by @_inception_ai labs' Mercury 2 model, which made the generated UI feel fast and responsive. To keep the output reliable, BayHop uses A2UI JSON with a registered widget catalog, so the model can only compose Flutter components the app actually knows how to render. That foundation is combined with real-time BART/511 transit data, Google Places, location context, and a custom BayHop design system to turn prompts like “Downtown Berkeley → SFO, leave now” into an interactive transit experience.

Huge thank you to @VGVentures and Flutter SF for hosting, mentoring, and bringing together such an awesome Flutter community.

Check out the project now at https://t.co/TXauWdJ1KW
Code: https://t.co/LLpX6VL12c

#Flutter #GenUI #GenerativeUI #AI #Hackathon #OpenSource #FlutterDev #InceptionLabs

[lfpmb @lfpmb1]

I built Rewind because some of the most important moments are the ones you only realize mattered after they already happened.

Rewind is an iPhone app that keeps a short, private, on-device rolling audio buffer. When something worth saving happens you press once and Rewind saves the last few minutes.

The clip is then transcribed, summarized, titled, and made searchable directly on your phone.

A few things I cared a lot about while building it:
• No cloud upload path
• No account required
• Local transcription and summaries
• Action Button, Siri, Shortcuts, and Spotlight support
• A rolling buffer that overwrites anything you don’t explicitly save

The goal is simple: let your phone remember the last few minutes for you, while you decide what is actually worth keeping.

https://t.co/S7PJZWBrKN

Would love for people to try it and share feedback.

[lfpmb @lfpmb1]

Rewind lets your iPhone save the last few minutes of audio after the moment already happened.

Action Button, Siri, Shortcuts, local transcription, no cloud, no account.

For the ideas, stories, and details you only realize mattered once they’re gone.

https://t.co/Hhc4LDLtRP

[lfpmb @lfpmb1]

Been working on this for a bit, a geography game with multiple modes:

https://t.co/BV2z446kqJ

[lfpmb @lfpmb1]

@heyblake https://t.co/JKMfE7N7QD

[lfpmb @lfpmb1]

How well do @Kalshi mention markets actually price probability?

I’ve been exploring mention markets across categories and wrote up a piece on calibration, mispricing, and the patterns that keep showing up in the data:

https://t.co/VdSApiu5UY

I have refined my methodology and approach analyzing mention markets. To make things more digestible, I've broken down the exploration into smaller articles, first we will be looking at general mention markets and analyzing trends within them.

[lfpmb @lfpmb1]

@AbhiCodes15 Java

[lfpmb @lfpmb1]

@KaiXCreator understanding

[lfpmb @lfpmb1]

@vineerpasam only trivial changes

[lfpmb @lfpmb1]

lets give this app a honest try for a month

[lfpmb @lfpmb1]

@0xReflection We are stuck in time

[lfpmb @lfpmb1]

@Mrcryptoxwhale What a guy

[lfpmb @lfpmb1]

@CoinvoTrading If I ever need a cherry picking data example i’ll use this video

[lfpmb @lfpmb1]

@Cryptoze bro just drew some lines and that’s it

[lfpmb @lfpmb1]

@MerlijnTrader do you know the equation for a parabola?

[lfpmb @lfpmb1]

@soapweb3 full send
https://t.co/f8J2m68FeZ