Case 03 · In progress
In progress: an AI content pipeline for solo creators
A content pipeline for solo creators: plan, generate and publish text, image and video from one workflow, with Claude and Fal behind it. In progress, no launch date promised.
Screenshots coming once the app store listing is public.
What I did
- Product scoping and interface design
- Next.js 16 app with Supabase auth and Postgres
- Claude and Fal orchestration
- Billing, email and media storage
- Launch checklist: monitoring and payment webhooks
Overview
Creators repeat the same loop: outline, draft, render, publish. This app turns that loop into a saved pipeline where text comes from Claude, images and video from Fal, and each run is versioned so a step can be re-run or rolled back.
I am building it alone as a product of my own, from the first sketch to the billing webhooks. It is not launched, and the case will grow figures once it is.
Product and client names are withheld under NDA. The stack, the architecture and the figures shown are accurate.
Stack
- Next.js 16
- TypeScript
- Supabase
- Anthropic SDK
- Fal
- S3 / CloudFront
Architecture
- Next.js 16 App Router
- Supabase Auth
- PostgreSQL
- Claude
- Fal (image / video)
- S3 / CloudFront
Decisions that mattered
- 01
Two model providers, one plan
Text steps call Claude, media steps call Fal, and a plan object decides what runs in sequence and what runs in parallel.
- 02
Token budgets per tier
Each account carries a token quota. Overspending downgrades the model for the rest of the cycle instead of failing the run.
- 03
Every run is a version
Pipeline runs are stored with their inputs, so a creator can regenerate one step or roll the whole post back.
- 04
Operations before invites
Error tracking, analytics and payment webhooks are wired before the first invite goes out.
More work
Have something to build?
Send me two paragraphs about it. I reply within 24 hours, Seoul time, with a price and a date.
Email me