AI adoption, proven by building
Everything I propose is something I build and use myself: an App Store app running a speech AI model I trained, a Slack bot that answers database questions in plain language, a metrics report that lands every morning. All in daily use, today.
Running right nowLifeVocab · Daily Wave · Daily Merge · Why Next Blog
Evidence
Everything I propose, I built and still run
No PoCs. No demos. These are live in the stores, with real users showing up every day.
LifeVocab
Live on the App StoreSnap a photo and AI turns what it sees into sentences you can use right away. Idea, model training, AI integration, store release: all done solo.
- On-device speech recognition
- Multimodal pipeline
- 5 languages
- Daily Wave
A mini game: ride today's wave, once a day. A new wave rolls in at midnight.
Check it out → - Why Next Blog
Field notes from running AI agents in parallel with cross-verification. How I work, out in the open.
Check it out → - Daily Merge
A 2048-style merge puzzle with the same board for everyone each day. Chase a high score in 30 moves, or take on the bot in the Arena.
Check it out → - Daily WordIn progress
One Korean word a day. Guess it. Coming soon as a Toss mini app.
How I can help
Let's decide where AI goes first
Skip the grand transformation. We find one thing that pays off within two weeks and actually run it.
Workflow automation
A Slack bot that queries your database in plain language, and metrics and error reports every morning. Hand the repetitive work to AI first.
AI inside your product
Speech models trained and shipped on-device, image generation, multimodal pipelines. Proven in live services on the stores, not PoCs.
AI adoption for dev teams
In AI coding, verification beats speed. Code review bots, guardrails, verification gates: workflows your team will actually use.
A roadmap that starts small
Diagnose, pilot, scale, in that order. We widen the scope only after you have seen the first win with your own eyes.
Please note: consulting sessions and deliverables are conducted in Korean.
Recent posts
View all postsOne Bucket, Two Terraform Owners - the Last apply Wins
I added one S3 lifecycle rule and terraform apply failed. The cause: two resources each owning the lifecycle configuration of the same bucket. S3 lifecycle is document-level, not rule-level, so the last winner silently erases the other side's rules.
The Tests Were Green - the Notification Had Never Been Sent, Not Once
The code that enqueued jobs had been failing every time since day one. The error was swallowed, and the tests stayed green thanks to a mock. The story of how a single colon quietly killed three features.
We Chased the 9 People Intercepting Our App's TLS. It Was Google.
We turned certificate pinning on in report-only mode and 74 SPKI mismatches piled up in Sentry. It looked like a MITM attack, and the AI firmly ruled out store review, saying 'Google does not intercept your HTTPS.' The real IPs pulled from our production API logs overturned that guess.
About the maker
Hi, I'm Why Next, and I build small apps on my own. I believe in shipping over grand plans, and in steady iteration over one big swing.
I consult the same way. Instead of borrowing someone else's case study, I only talk about what I have run into myself.
Contact
What would you like to automate?
A line or two is plenty: the work that keeps repeating, or the AI feature you want to ship. I'll read it and reply by email.
If email is easier, reach me at admin@whynext.app