Sngular

The AI work here that carries both a named client and a published number is Connected Works, built with Ferrovial's Digital Hub: a magnet-mounted IoT unit with GPS, accelerometer and gyroscope feeding a vibration-analysis algorithm that classifies loading, unloading and transport without operator input, running on the Tiétar high-speed rail works and High Speed Two and reporting up to 20% less machinery wait time.

The generative-AI delivery sits on the services page instead, and none of it is attributable: a financial-sector client's platform for managing LLM, speech and multimodal model lifecycles—fine-tuning, guardrails, quota-controlled inference endpoints, vector stores, agent creation with an SDK—plus a tender-screening explorer and a scholarship assistant for a platform covering more than 1,000 universities, each written up without a client name or an outcome figure.

Of the 20 published success stories, most are Atlassian and Jira Service Management rollouts, mobile and cloud builds—B100 for ABANCA runs a Java backend on Kubernetes on Google Cloud with a BigQuery customer data platform—and museum installations; the closest thing to an automation number is Vítaly, where 17 processes moved onto three RPA robots and customer-service response times fell 60%.

The weaker fit is buyers who need to read a named, measured language-model deployment before they shortlist—here the LLM work and the named clients sit in different documents, so ask for one reference that covers both.