Case study 16
StartDescribe
StartDescribe turns product listings into high-converting, multilingual, SEO-ready pages for dropshippers, retailers, multilingual sellers and agencies. Merchants install it in Shopify, connect WooCommerce or BigCommerce, or upload a CSV or Google Sheet; StartDescribe fetches the products, and an AI pipeline writes descriptions, bullet points, pros and cons and SEO copy, enhances images and translates the results. After review and approval, the content syncs back to the store, and new products are enhanced automatically as they arrive. Teams pay through tiered Stripe subscriptions. Reworked onboarding and automation raised onboarding conversion by 40% and cut content generation time by 30%.
- Sole engineer
- Client Project
- AI, SaaS
- Generative AI, E-commerce, Shopify, SEO, Multilingual Content, Subscription Billing
- Next.js
- TypeScript
- Tailwind
- Supabase
- Shopify
- Stripe
- OpenAI
- Perplexity
What StartDescribe does
- Install as a Shopify app, or connect WooCommerce or BigCommerce
- Bulk import from CSV or Google Sheets, with field mapping
- Automatic product fetching, and real-time enhancement of new products
- AI-written descriptions, bullet points, pros and cons, and SEO copy
- AI image enhancement
- Multilingual content with automatic translation
- Review and approval before publishing, then sync back to the store
- Team permissions, a client dashboard and white-label options for agencies
- Zapier, Make and n8n modules, plus a REST API
- Tiered Stripe subscriptions with a free trial
- Transactional emails
The problem
Store owners with hundreds of products need descriptions and images that are accurate, on-brand and good for search, without spending a day per catalogue.
Approach
A generation pipeline first researches the product, then drafts copy and images with the store's tone and constraints in the prompt. Products are fetched from Shopify automatically, and results land in the merchant's workspace for review before they sync back to the store. Onboarding was shortened so a new store sees its first generated description quickly.
Key decisions
- D1
Research before writing
A retrieval step gathers product facts first, so the model writes from context instead of guessing.
- D2
Human in the loop
Generated content stays a draft until the merchant approves it.
- D3
Managed foundations
Supabase for auth and Postgres, Stripe for billing: engineering time goes into the pipeline, not the plumbing.
Highlights
- Built the Shopify app. Merchants install it in their store, products are fetched automatically, and AI-enhanced descriptions and images sync back to Shopify.
- Implemented secure authentication, an AI content generation pipeline, transactional emails and SEO-optimized pages.
- Increased onboarding conversion by 40% and reduced content generation time by 30% through optimized workflows and automation.
My contributions
- Turned the client's requirements into a product scope and architecture
- Built the whole product alone: Next.js frontend and server routes, Supabase auth and database, and Tailwind UI
- Built the Shopify app: installation, automatic product fetching and syncing enhanced content back to the store
- Built the WooCommerce and BigCommerce connections and the CSV and Google Sheets import
- Built the AI content generation pipeline on OpenAI and Perplexity
- Integrated Stripe subscriptions and transactional emails
- Reworked onboarding and automated workflows, raising onboarding conversion by 40% and cutting content generation time by 30%