Fable 5 - biggest leap so far when it comes to LLMs? My experience: for the 3 days: 2 engineer-months of senior iOS work. 4 App Store releases. Test coverage up 64%. A confirmed fix for a crash I couldn’t reproduce.
All in 3 calendar days, directing Fable 5.
Launched a new site for the health app I’ve built, it compares you against your own data, calculatea bio age, and nudges you towards healthier habits. https://t.co/GEu83Sr3bJ
Building with Claude and having Codex to verify is the best, and sometimes also I have Codex to do the bug fixes.
I am not even sure its really about which is better, but for some reason I like interacting with Claude more. Today I built a better Fitness Score into Kiito.
One key differentiator is as well that everything is calculated on-device, so no data leaves your phone, and when AI is used it’s the Apple’s on-device AI.
Link to app: https://t.co/OgQJYSuu4Y
I built a new app called Drift: Longevity Habits for iPhone, and it has two core features:
1. Drift score which gives you a day to day alignment against your historical data
2. Longevity score with bio age - how are you doing vs. 8 key longevity metrics
#ios#longevity
Into iOS app development lately, and got my first two apps released on the app store this week
I made three design decissions for all my apps:
1. No backend - everything on-device
2. No third party dependencies - nothing to patch
3. Use new features of Apple SDK
CISO’s + LLM = great at building? Built an Azure Lab utilizing the the Startup-Scale Landing Zone as a basis for the plan, but still making a Lab Environment to fit to purpose. Great experience, also used the new Azure Skills for building + Drawio skill for making this drawing.
In the past 37 days I’ve been learning on how these new agentic coding tools (OpenClaw, Claude Code & Codex) work. So much so that having build 20+ solutions, I already feel that I am a builder first and CISO second, even though my first motivation to learning was security.
As a new evening cron job, before brushing your teeth, I suggest asking your OpenClaw and other agentic AI’s and LLM’s what they know about you and then make the adjustments that make you more comfortable in using them tomorrow. 3/3
Opsec and Openclaw? So I've been playing around with OpenClaw on a dedicated machine that has no personal information on it, and I am trying to hide who I am from Openclaw. Which is not nontrivial - when you want to communicate it with somehow. 1/3
@JussiRoine How much weight? I saw this video recently in youtube where someone had some students do some weighted pull-ups to see how it correlates to muscle-up. I think the result was that if you can do a pull-up with 54% body weight you can potentially do a muscle-up.
@WorTony@andresbohren It does, high secure score indicates for the attacker it is not as likely get return on investment for attacking such an environment. This is very useful, as the attacker can instead direct their efforts elsewhere with a more likely ROI.