KG.codes

I would never build a mobile app without using these 3 free services.

1. OneSignal.
It gives you free push notifications inside your app.
Push notifications are critical for getting people back into your app after they’ve left, letting them know what’s happening, and driving usage when you need it.

2. RevenueCat.
If you’re trying to make money from your app, sell anything through the App Store, or add subscriptions, RevenueCat makes it super simple.
Some people like Superwall, which is great too, but I prefer RevenueCat.
You can set up paywalls and subscriptions easily and quickly, and it works across iOS and Android.

3. Sentry.
You might say this ain’t super critical, but you need to know when your app is crashing.
You need to know how the app crashed, what the user was doing, and the deeper technical context + stack traces.
You also need to detect usage patterns that lead to crashes so you can fix issues and improve the app.

Thoughts?

Drop any other free services you’d add.

People keep saying AI can already replace software engineers.

But then when an exec (like the COO from Uber) says they’re not seeing the ROI from AI yet, the response is suddenly, “they didn’t train people enough” or “they’re not using it right.”

So which one is it? If humans can already be replaced, why do humans need to get better at using AI first?

What do y’all think?

AI should be testing its own work at the very least.

Codex (ChatGPT) is getting really good, specifically the macOS app. I’ve got it testing my mobile app inside the simulator, controlling the screen, validating its own work, and fixing issues without me having to manually point everything out.

Claude/Opus might still more intelligent at peak but the value versus token burn feels better with GPT.

People are starting to use AI to commit refund f*aud through DoorDash.

For example, they upload a picture of their food to ChatGPT and say, “Make it look undercooked,” then use that generated image to try to get refunds.

This was possible before of course but the AI tools today make it a lot easier/faster to do, and harder to get caught.

I’m sure to many it might seem smart or funny when people are just scamming big companies and getting their money back.

But what happens when people start turning this stuff on each other?

What happens when you’re trying to sell something on Facebook Marketplace or eBay and somebody edits a picture to make it seem broken or damaged just to ask for a partial refund?

Or you’re in the service business, you finish a job at somebody’s house, leave, and then they edit a picture afterward and say the work wasn’t done right.

Even if you can go back and check, that still costs time and money.

I feel like we’re getting to a point where picture verification alone can’t really be trusted anymore.

We’re going to have to find new ways to verify things. Trust, but verify.

What do y’all think about this whole thing?

Source video: