
Indie app developer/designer of @HoursTracker for iOS and Google Play.
TX
Joined July 2009
- Tweets15.2K
- Following211
- Followers863
- Likes2.5K
Carlos Ribas retweeted
if your favorite ASO guru tells you not to repeat keywords on App Store
get a new one
Replying to @IndieBeaverHere
i’m not joking.. the reason is actually two-fold:
1. you want your top keywords in the keyword field to make CPPs for them
2. the App Store algo doesn’t punish duplicates the same way anymore, as long as you don’t put ’habit tracker’ in both title and subtle—instead you should put it in the title then ’habit tracking’ in the subtitle
Carlos Ribas retweeted
Got an iPhone Duo?
HoursTracker for iPhone Duo is currently available for beta testing. Sign up, give it a try!
testflight.apple.com/join/2J…
Carlos Ribas retweeted
Big Tech: We’re using AI to solve trillion dollar problems!
Me: Like what?
Big Tech: Wages.
Carlos Ribas retweeted
It's been a lot of fun trying the Duo out for a week, but it feels like I've gotten to use an extraordinarily polished prototype — not a finished product that I'd want to buy and carry.
daringfireball.net/2026/10/t…
Carlos Ribas retweeted
I’m sure this one won’t make anyone angry. (But probably not here.)
daringfireball.net/2026/10/x…
Carlos Ribas retweeted
Building apps on demand is just way too much cognitive overhead for the average person. No one wants to do this.
If I go to Airbnb, it loads in a second. Designers have thought through every detail. They control the data. They handle the payment processing. If something goes wrong with my stay, I know who to call.
I don't want to have to write up a description of what I want to do for every little task, and have a bespoke, throw-away interface created for me every time I need it. I'm going to have to learn how to use a different piece of software every day. It's going to be way worse than what Airbnb made, and I'll have to wait for it to generate. Didn't have something I wanted? Sure I can prompt it, but then I need to wait again. Honestly, it just seems like a massive annoyance.
I just want my interfaces to be stable, well-maintained, and designed with care. I really don't care that I can generate it if I want to. It's just not practical. And unless inference can generate a site better than Airbnb in a second, I just am not going to use it.
This is one of those things that is "cool" but in reality no one will do it.
For example, look at how many people still use Google search, 3.5 years after ChatGPT was launched. People said search was going to die. The reality is that search volume has never been higher.
I think it'll be more like:
You chat to your LLM app and it'll just spin up an app with a UI if it feels the need
Like you want to book a holiday apartment, it spins up a fancy Airbnb-like interface and books it for you
You want to edit your photo, it spins up a simple but advanced easy-to-use photo editor
The idea is that interfaces and apps become ephemeral to help you achieve whatever you want in a moment
everyone's talking about their teams like they were at the peak of efficiency and bottlenecked by ability to produce code
here's what things actually look like
- your org rarely has good ideas. ideas being expensive to implement was actually helping
- majority of workers have no reason to be super motivated, they want to do their 9-5 and get back to their life
- they're not using AI to be 10x more effective they're using it to churn out their tasks with less energy spend
- the 2 people on your team that actually tried are now flattened by the slop code everyone is producing, they will quit soon
- even when you produce work faster you're still bottlenecked by bureaucracy and the dozen other realities of shipping something real
- your CFO is like what do you mean each engineer now costs $2000 extra per month in LLM bills
Carlos Ribas retweeted
Are the Moltbook posts fake? Yes.
Are the Moltbook posts real? Yes.
How are you defining real or fake here?
"An LLM wrote them" - yes that's literally the point
"Humans set them up to do this" - yes that was always acknowledged
"They're just mimicking Reddit posts" - yes
Carlos Ribas retweeted
Update: the foundation models will eat every possible business on top of them
If software is cheap to build & the foundation model is the true scarce asset, why wouldn't they do every application?
Carlos Ribas retweeted
Replying to @DaveShapi
But inference will not be free, so all software revenue will rapidly divert to whoever provides inference at scale.
Carlos Ribas retweeted
📣Announcing the first preview releases of Swift for Android, enabling you to build Android business logic with the same Swift that you use for Apple platforms. swift.org/blog/nightly-swift… #Android
Carlos Ribas retweeted
Developer kept complaining our RDS instance was too slow. Wanted to upgrade from db.t3.large to db.r5.8xlarge.
Cost increase: $140/month to $3,400/month.
I checked query performance first:
- Query ran 47,000 times per hour
- No indexes on the WHERE clause columns
- Missing composite index cost 890ms per query
Added two indexes. Query time: 890ms -> 12ms.
Saved $39,000 annually because we spent 15 minutes with EXPLAINING.
Most performance problems are code problems disguised as infrastructure problems.
Carlos Ribas retweeted
You all understand that the AI bubble is about to pop, right? And when it does all the hiring managers, who have been holding their breath waiting for the AI miracle, will take a deep breath, realize in horror that they are badly understaffed, and start hiring programmers with reckless abandon.
Carlos Ribas retweeted
I sent this note to our entire team at @Opendoor earlier today.
Carlos Ribas retweeted
As I get more experienced with using LLM assistance, I'm finding it closer to supercharged autocomplete than anything else. And I get better results if I think about it that way too.
If I simply write prompt after prompt without the closest idea of what it's actually doing, things inevitably implode.
Not to mention, costs rack up fast. If you just wantonly shoot for GPT-5 and Opus, then your experience is not going to be cheap.
Finally, when AI writes anything complicated, there's a ~80% chance I have to throw it away completely or rewrite it.
Full-automated assistance is probably in our future, but it's definitely not happening anytime soon.
Carlos Ribas retweeted
New icon mask for iOS/iPad/macOS 26 is a rounder
parametric squircle. #WWDC25
Carlos Ribas retweeted
Long-term, AI is not replacing any software engineers, and in fact will lead to the wildest headcount growth this industry has ever seen.
Yes, there are leaders out there who are experts in value extraction over revenue growth who are massively reducing engineering headcount right now (free money, right?). Hate to break it to them, but this is incredibly shortsighted thinking.
Everyone else also got the memo too—yeah, it's no secret that your engineers can now get more done faster.
Anyways, the folks who laid of 50% of their team because "look at me, I can get the same work with 1 engineer instead of 2!", will, in a few months, start to notice "something's off" with churn. A few months later, they'll notice that ARR is actually decreasing. And then, they'll discover that ✨ surprise ✨, their biggest competitor decided NOT to lay off 50% of their team, and instead went all in on training everyone to 10x their productive time in seat, and now is running circles around them.
OOPS.
Sorry, you messed up. And you likely won't catch up.
Yes, the detractors keep pointing to charts showing the massive decline in job postings for software engineering roles post 2022. "This is just the beginning of a long-term trend", they say. And you know, it could be how the next 12-24 months go, depending on the economic fallout from these tariffs or a recession. But it's nothing to do with AI.
Macroeconomic conditions aside, my prediction is that over the next 10-15 years, software engineering as a profession is about to see the wildest growth in headcount that we'll have ever seen. And it's because of a combination of productivity levers, reduction to cost-of-entry, and wild growth in demand for software that just couldn't be built prior to 2024.
Carlos Ribas retweeted
Replying to @dwlz
Yes. When power tools were invented (circular saws, drills), builders weren't out of work, they could just build houses faster. But the important takeaway is once electric tools were invented, the builder who opted out was quickly out-competed by those who embraced the new tools.
Carlos Ribas retweeted
Mark Rober: Should we show how Luminar LIDAR based cars perform much better in dense urban traffic?
Luminar: No, our system can't do that
Mark Rober: Should we show how much better it is than Tesla at driverlessly operating in parking lots
Luminar: No, our system can't do that.
Mark Rober: Well, what is your system better at?
Luminar: Well, you know how in Road Runner Will E. Coyote would always drive into the wall because roadrunner painted a road on it? Well, our system should be able to perform better on that if we compare it to a five year old version of Autopilot. We can't use FSD as the latest occupancy network would probably detect it, just like how a human can see it with their eyes
Mark Rober: Oh yeah, that happens to me all the time. That's much more important to your average driver than those first few ideas I had
























