Skip to content
Sectors
Sectors · Logistics

AI in supply chain, traceable and audit-defensible on SLAs.

AI consulting for logistics, supply chain and last-mile. Demand forecasting, optimization, operations assistance. Traceable approach, aligned with contractual constraints (SLAs, penalties, KPIs).

Warehouse dock at night: forklift silhouette, warm sodium light from the doorway, concrete floor
In 3 sentences

Colombani.ai supports logistics and supply chain actors in AI integration: demand forecasting, last-mile route optimization, warehouse operations assistance, KPI analysis. Traceable approach required by your customers (SLAs, penalties), GDPR-compliant on driver data and CNIL geolocation deliberation.

Use cases

SKU-level demand forecasting

Short and medium-term forecasts by SKU and site, integrating seasonality, promotions and market signals. Forecast explainability for the planning team.

Last-mile route optimization

Multi-constraint optimization (delivery windows, vehicle capacity, driver time). Real-time adaptation to incidents.

Warehouse manager / operations assistant

Morning route preparation support, alerts on plan deviations, end-of-shift summary. Traceable for quality audit.

Ethics and regulatory framework

GDPR — driver data and geolocation

CNIL 2015 deliberation frames employee geolocation. Legal basis, information, retention duration, limited access. Colombani.ai adapts the architecture to respect this framework.

Contractual SLA and penalty constraints

Your customers require SLAs on deliveries. Any AI use must be traceable to be defensible in case of dispute. Colombani.ai delivers expected logs and reports.

What we don't do

  • Physical network optimization (warehouse siting): other expertise.
  • Vehicle embedded systems: other expertise and specific regulatory framework.
  • Capacity trading (freight markets): financial framework, not covered.

Frequently asked questions

How do you integrate with an existing TMS? +

Via a dedicated MCP connector for your TMS (Transwide, DDS, MEA, Descartes, or custom). The agent reads and writes in your TMS via API, without data copies elsewhere. Typical development: 2 to 4 weeks for a simple connector.

Discuss your project

First call is free. Straight answer on what is feasible and what is not within your ethics framework.