← All case studies
Case study
LuxenCrest — digital content agency
An end-to-end AI editorial pipeline — from trend monitoring to multi-platform publishing, with no manual intervention on generation.
Delivered by AI Docking
- The luxencrest.com website
- The editorial & communications agency application (the agentic backend below)
Context
LuxenCrest is a digital content agency publishing on Instagram, LinkedIn and Twitter/X. The content — copy, visuals, carousels — is entirely produced by a system of agents, not an editorial team.
What was built
LangGraph agentic backend (7 agents: trend analyst, editorial direction, strategist, writing, image generation, QA, publishing), FastAPI + PostgreSQL + Redis, deployed on a VPS with observability (Arize Phoenix).
- LangGraph — 7 agents (veille, direction éditoriale, stratège, rédaction, image, QA, publication)
- FastAPI + PostgreSQL 16 + Redis 7 (checkpointer LangGraph)
- Publication Instagram Graph API, LinkedIn, Twitter/X
- Observabilité Arize Phoenix (durée, tokens, seuils par nœud)
Result
- 20+ campaigns produced, blending copy and images
- Under 2 minutes to get an Instagram campaign ready to approve or iterate on
- Rigorous token-consumption tracking, in the client's own hands
Figures extracted from the production database, provided by the client.