2 min read
Elevate chose its first vertical — we’re doubling down on content for now
First vertical is 가게점수, built as a separate product. Elevate shifts to blog, newsletter, and ops — a stronger commit to media-first (ADR-012).
Earlier this year we kept asking “what ships next.” Today the public surface is still mostly waitlist + auth, plus the content and automation backbone we already run. Shipping more app features before the story earns attention felt backwards. We’d rather publish writing worth reading — consistent with the media-first decision (ADR-012, in-repo).
Our first vertical is 가게점수 — a marketing diagnosis assistant for solo café, dessert, and bakery owners: paste shop context, get a score and an action plan. It is not an Elevate feature. New domain, new repo, new brand, no code imported from here. Pricing, spec, and demos belong on 가게점수’s own channels; this post is only the direction share for the Elevate audience.
Why that wedge: several “desire-axis” consumer verticals in Korea already have dominant players who have moved fast on AI; pitching there means burning budget on headwind. The gap we’re testing first is closer to SMB / self-employed AI assistance, narrowed to single-owner hospitality retail. Deeper market notes will live in 가게점수 docs/decisions/.
What Elevate does next (~3 months): pause new product UI/features, new marketing pages, and cosmetic redesigns unless safety demands it. We ship blog posts 3×/week (Mon/Wed/Fri), newsletter 1×/week (Thursday), grow SEO + email, and keep INIT/STAB work (incidents, deps, PostHog). No 가게점수 implementation in this repo.
You’ll see builder-facing insights here — workflows, prompts, SaaS ops, founder notes — rotating topics so this doesn’t turn into a small-business-only feed (that ICP is 가게점수).
Takeaway: Follow along via the blog and newsletter. Join the waitlist → if you want the quiet ping. Launch details for 가게점수 will come separately when that product is ready to speak for itself.
Continue reading
- 9 min
If you cannot measure your agent loop, you cannot improve it
A weekly agent ops scorecard helps teams improve AI workflows with simple metrics, reproducible reviews, and fewer model-hype detours.
- 9 min
Build the funnel before the feature
I almost committed a new feature ADR yesterday. A two-minute analytics query stopped me. Here is what the data actually said and the rule I now apply before any next build.
- 4 min
The mechanism works. No one knows.
Five-minute audit: my payment system worked, my checkout worked, my essays were published. Outside customers: zero. What I missed about L4.