Apple's Gatekeepers vs. A Rubber Duck
The App Store's black box is a nightmare. This is the cheat sheet.
We spend so much time talking about AI building new things, we forget it can also fix broken systems. And the App Store review process is definitely broken.
This Tool Catches App Store Rejections Before Apple Does
A new service combines automated checks and human testers to predict why Apple's reviewers will say no.
Rubber Duck is a pre-submission scanner for iOS apps that hunts for the exact issues that get you rejected. It checks for common failures like crashes, broken links, or missing privacy details that automated systems often miss. The clever part is that it combines bots with real human testers, delivering a report in hours.
This is not just a linter; it is a direct response to the industry's quiet acceptance of Apple's opaque review process. Developers have been conditioned to treat submission as a lottery, wasting countless hours on preventable rejections. The real story is that we are now building services to navigate arbitrary gatekeepers, treating a platform's flaws as a solvable engineering problem.
For indie developers and small teams, this provides a massive advantage, saving weeks of back-and-forth and turning a gamble into a predictable process. But the bigger question is what it says about a platform when an entire cottage industry emerges just to guess what its rules are.
The New Creative Co-pilots
AI is moving past novelty tricks and into the timeline, augmenting workflows instead of just generating images.
FireCut for DaVinci Resolve: Your new AI assistant editor
FireCut automates the most tedious parts of video editing, like cutting silences and adding zooms, directly within DaVinci Resolve. This signals a shift from editing as a manual task to a process of creative direction, where you review the AI's grunt work, not do it yourself.
DemoStudio: A film studio for your product demos, in a browser
This gives everyone powerful, browser-based demo creation tools with 4K export for free. It completely erases the barrier to entry for creating polished product showcases, levelling the playing field for indie makers against big-budget marketing teams.
Questas: AI-powered choose-your-own-adventure stories
Questas generates visual, branching narratives from a simple prompt, turning anyone into a storyteller. It is less about a finished product and more a glimpse into a future where interactive entertainment is cheap and easy to produce.
Building Smarter Plumbing
As the AI hype cycle cools, the real work is in building the infrastructure to make it reliable and performant.
AskCodi: An orchestration layer for your LLMs
AskCodi helps developers manage and build with custom LLMs without the pain of retraining models. This is the unglamorous but critical work required to move from AI experiments to production-ready applications that do not rely on a single vendor.
SlopCollector: A raccoon to clean up your Supabase mess
This tool focuses entirely on optimising Supabase databases by finding missing indexes and slow queries. It’s a perfect example of a new wave of niche, developer-focused tools that solve one painful problem really well.
Quick hits
Convo: Your AI meeting assistant
Convo automates meeting prep, offers live suggestions, and writes your follow-up emails, aiming to kill meeting fatigue for good.
InterviewFlowAI: The AI-powered hiring filter
This automates first-round interviews, from resume scoring to AI-led calls, freeing up recruiters from the soul-crushing part of their job.
Orion: A browser that minds its own business
Built on WebKit, Orion delivers Safari's speed with zero telemetry and full support for Chrome and Firefox extensions.
My takeaway
AI is getting boring, and that's the best thing that could happen to it.
The era of magical, world-changing demos is giving way to a wave of practical, problem-solving tools that feel more like plugins than revolutions. This maturity means the technology is finally becoming useful infrastructure instead of just a spectacle. The focus is shifting from 'what can it do?' to 'what can it fix?'.
This means we will see more co-pilots, assistants, and optimisers that slot into existing workflows. The real innovation is moving from the foundation models to the clever application layer that targets niche, painful problems. The next decade will not be about one giant AI, but thousands of tiny ones doing the boring work perfectly.
What tedious part of your job are you just waiting for an AI to solve?
Drop me a reply. Till next time, this is Louis, and you are reading Louis.log().