Composable commerce (commercetools, Contentful, Algolia, Adyen) gave us flexible storefronts. AI is now the personalization layer on top.
Stack I ship
- Next.js storefronts with ISR for SEO
- commercetools for catalog, cart, B2B business units
- Algolia / Coveo for search + merchandising rules
- LLM layer for recommendations, support copy, and social commerce intents
AI use cases that actually ship
| Use case | Approach |
|---|---|
| PLP ranking hints | Rules + ML scores, not pure LLM |
| Support / order chat | LangGraph + commerce API tools |
| Content generation | Constrained templates + human review |
| Social selling | Structured intents first, LLM second |
Social commerce connection
B2B distributors don't want a generic chatbot — they want order numbers, proposals, and approvals wired to commercetools. That's social commerce, not gimmicks.
Takeaway
AI in commerce wins when it respects APIs, scopes, and audit trails — same discipline as payments and PCI.
