// ai supply chain · UAE · GCC · US markets

Chargebacks won.
Exceptions triaged.
Operations automated.

Two AI copilots sharing one intelligence layer. Compliance Shield recovers lost revenue from OTIF chargebacks. Neural Command routes exceptions before the morning standup.

5–15%
Chargeback recovery lift
20–40%
Exception resolution time ↓
95%+
Infrastructure margin
$635/mo
Infra at 5 clients
Two products · one shared brain

Pick your problem.

Both prototypes run on PostgreSQL + pgvector. One shared Claude intelligence layer. Deploy together or independently.

Prototype 1

Compliance Shield

For supply chain teams losing chargeback disputes they could have won. AI-drafted dispute letters grounded in routing guide evidence. Deterministic OTIF engine. Human review before any submission.

  • Real-time OTIF failure detection against retailer SLAs
  • pgvector RAG — routing guide clause retrieval
  • Claude drafts dispute with evidence_chain (no hallucinations)
  • Chargeback state machine: new → disputed → recovered
  • Human review gate — always in the loop
Open Dashboard →
Prototype 2

Neural Command

For ops teams drowning in exception noise from 5 source systems. Normalised, scored, routed. TriageAgent every 15 min. S&OP brief written before standup.

  • 5-source event ingestion (TMS, WMS, ERP, Telematics, EDI)
  • PriorityScorer: revenue × tier × penalty × urgency
  • TriageAgent routes top-20 exceptions every 15 min
  • SummaryAgent writes daily S&OP brief via Claude
  • Slack / Teams notifications with idempotency
Open Console →
Shared platform stack

One codebase. Two products.

FastAPI (Python 3.11) Next.js 15 PostgreSQL 16 pgvector Claude API (Anthropic) MCP Tool-Calling Celery + Redis Docker + ECS AWS ME-South-1 (UAE) AWS US-East-1 Supabase (dev) JWT + Okta SSO Alembic migrations APScheduler