{"@context":"https://schema.org","@type":"HowTo","@id":"https://bkamp.ai/de/recipes/feb03b81-b463-4cae-b2ec-2201848b20bc","name":"[WIV_SG] HelperHub – ein KI-generierter Speiseplan für Haushalte mit Haushaltshilfen","description":"Eine KI-gestützte Haushaltsmanagement-App für berufstätige Mütter mit im Haushalt lebenden Haushaltshilfen – entwickelt als Mobile-First PWA.","url":"https://bkamp.ai/de/recipes/feb03b81-b463-4cae-b2ec-2201848b20bc","image":"https://bkamp.ai/bkamp_home.png","inLanguage":"de-DE","datePublished":"2026-04-09T03:12:18.493818Z","dateModified":"2026-04-09T03:12:18.492372Z","author":{"@type":"Person","name":"yoonsarah_4b4ccc","url":"https://bkamp.ai/de/users/yoonsarah_4b4ccc"},"publisher":{"@type":"Organization","name":"bkamp","url":"https://bkamp.ai"},"educationalLevel":"초급","category":"개발","step":[{"@type":"HowToStep","position":1,"name":"Why I built this","text":"

Why I built this: Weekly meal planning is repetitive and tedious, but starting from scratch every week is even more exhausting. I wanted to automate the most time-consuming part of running a household with a helper — so I could set it once and let it run.

"},{"@type":"HowToStep","position":2,"name":"Key features","text":"

Phase 1 — Now

AI Meal Planner

Target: Korean working moms in Singapore & Hong Kong

✓7-day AI meal plan generation

✓Step-by-step recipes (Grade 3 English)

✓Lunchbox & snackbox generation

✓Employer ↔ Helper invite system

✓Supabase auth + RLS

✓Deployed on Vercel

○Grocery list

○Helper view + Approve & send

○Cooking log

Phase 2 — Next

Full Household OS

Target: All working moms with live-in helpers, globally

○Shared household calendar

○Daily task list for helper

○House rules documentation

○Helper profile & repertoire

○Favorites + AI weighting

○Multi-language support

○Baby/toddler separate meal plan

Goal: a complete system that works even when your helper changes — onboard a new helper in minutes, not days.

Service flow

Onboarding→ AI generates meal plan→Employer approves→Helper executes

Helper joins via one-time invite code · Role-based access · Supabase RLS scoped to household

"},{"@type":"HowToStep","position":3,"name":"Troubleshooting requested","text":"

Troubleshooting — feedback requested

1. YouTube recipe matching accuracy

Problem: Claude API guesses VIDEO_IDs — \"Korean egg toast\" matched a \"Korean street toast\" video (completely different dish). YouTube Data API has a 10,000 unit/day free quota, which runs out fast.

What I tried: Switched to free thumbnail URL pattern (img.youtube.com/vi/VIDEO_ID/...) + added strict dish-name matching rules in the prompt.

Where I'm stuck: Is there a scalable, cost-free way to reliably match a dish name to a YouTube video?

2. Supabase RLS recursive policy issue

Problem: The profiles table RLS policy referenced itself → infinite recursion. On signup, INSERT into households was also blocked because no profile row existed yet (auth.uid() returned null).

What I tried: Created a SECURITY DEFINER function get_user_household_id() to bypass RLS on the profiles lookup. Built create_employer_household() as an atomic RPC to handle both inserts in one transaction.

Where I'm stuck: Is this the right pattern for multi-table bootstrapping in Supabase, or is there a cleaner approach?

3. Parsing user-submitted recipe links (Instagram / YouTube)

Goal: Let users paste a recipe link, and have Claude rewrite it to match the helper's cooking level.

Problem: Instagram blocks scraping. YouTube requires transcript API or description parsing — unclear how reliable this is in production.

Where I'm stuck: What's a realistic, production-ready approach to extracting recipe content from social media links?

"}],"interactionStatistic":[{"@type":"InteractionCounter","interactionType":"https://schema.org/ViewAction","userInteractionCount":4},{"@type":"InteractionCounter","interactionType":"https://schema.org/BookmarkAction","userInteractionCount":0}],"mainEntityOfPage":{"@type":"WebPage","@id":"https://bkamp.ai/de/recipes/feb03b81-b463-4cae-b2ec-2201848b20bc"}}

[WIV_SG] HelperHub – ein KI-generierter Speiseplan für Haushalte mit Haushaltshilfen

Warum ich das gebaut habe

Hauptmerkmale

Fehlerbehebung angefordert

Step Roadmap

댓글 0개

[WIV_SG] HelperHub – ein KI-generierter Speiseplan für Haushalte mit Haushaltshilfen

Warum ich das gebaut habe

Hauptmerkmale

Fehlerbehebung angefordert

Step Roadmap

댓글 0개

[WIV_SG] HelperHub – ein KI-generierter Speiseplan für Haushalte mit Haushaltshilfen

Warum ich das gebaut habe

Hauptmerkmale

Fehlerbehebung angefordert

Step Roadmap

댓글 0개

[WIV_SG] HelperHub – ein KI-generierter Speiseplan für Haushalte mit Haushaltshilfen

Warum ich das gebaut habe

Hauptmerkmale

Fehlerbehebung angefordert

Step Roadmap

댓글 0개