Services
Systems that run, not strategy that sits.
Four ways I put content, marketing, and operations on autopilot with AI — each one designed, built, and operated end to end.
AI content & publishing automation
Source gathering, drafting, a quality gate, publishing, and mailing are wired into one pipeline that a scheduler runs with no human in the loop. The hard part isn't generation — it's reliability: when an upstream dependency (an LLM or mail API) fails, graceful fallbacks plus alerting and a rollback path prevent a 'zero posts today' outage. spoonai.me runs on exactly this design.
spoonai.me — a daily 08:00 KST cron collects, analyzes, drafts, and sends with no human touch. Weekday dailies + a Saturday weekly, every week.
Multi-channel publishing
Rewriting the same message per channel by hand is the work that quietly eats a week. One source is auto-adapted to each channel's SEO, hooks, and length rules, then published together — with canonical URLs to avoid duplicate-content penalties. Per-channel format, image, and tag strategy are fixed in code, so every run is consistent instead of improvised.
1 source → spoonai (KO + EN), DEV.to (English technical), and Naver (Korean local/compliance-aware) — four channel-native outputs, each formatted for that platform's ranking rules.
SEO & AEO optimization
Search is moving from 'ten blue links' to 'one answer.' Answer engines cite only 2–7 domains per response, so being citable matters more than ranking #4. I build for both at once: answer-first structure (question-led headings + a 40–60 word direct answer that gets lifted), Person/Organization entity markup with schema and sameAs, server-rendered HTML (most AI crawlers don't execute JavaScript), and correct hreflang/llms.txt — so search engines and answer engines surface and quote you.
AI answers cite ~2–7 sources vs Google's ~10 links. FAQ rich results were retired in May 2026, yet FAQ-structured content still aids AI extraction — this is the kind of current detail the build reflects, not last year's SEO.
Diagnostics & reporting automation
Diagnosis runs on measured evidence, not a hunch. Seven axes — local listing, blog SEO, AEO/GEO, paid search, website, reviews, and social/video — are each scored from browser-verified data and normalized to 100, surfacing in numbers where customers leak out (or never get in). Every score is backed by a captured screenshot, so nothing is estimated. The output is two artifacts: a client-facing HTML/PDF report built to convert, and an internal playbook I execute — sales and fulfillment in one engine.
A local clinic: 915 monthly visits to its map listing, but 0% from treatment keywords (≈80% were brand-name searches). Booking conversion was healthy (7 reservations/week) — so the data showed the funnel's top was empty: fill demand-side keywords and bookings follow.
This site is built to be cited by AI — not just found.
Explore the AEO build →Build it, ship it, let the results talk.
Start from the real funnel — what actually brings customers, measured from live data, not assumptions.
Build the smallest system that removes the most manual work, then harden it so it runs unattended.
Ship deployed, monitored results — not a proposal you have to act on yourself.
Contact
Let's build something with AI.
jd@jidonglab.comAI products, automation, integrations, and one-off builds — remote, worldwide.