Case Studies

Pets at Home, the UK's largest pet-care retailer, deployed a Microsoft Copilot Studio-powered AI agent for its profit protection team that cuts transaction analysis time from 30 minutes to under two minutes. The agent identifies fraud 10 times faster, enables the team to process 20 times more cases per day, and is projected to deliver seven-figure annual savings. Built in just 10 working days, it represents one of the earliest production deployments of Microsoft's agentic AI platform in UK retail.

MandarinMaster, a French EdTech startup, integrated IBM Watson Speech to Text to tackle the 80% dropout rate among Chinese language learners. The AI-powered platform enables real-time pronunciation validation, simulating a private tutor experience and helping absolute beginners reach C1 fluency in three years through daily 30-minute lessons where learners speak 80% of the time.

Swiggy, India's leading food delivery platform processing over 5 million daily orders, deployed a Multi-Input Multi-Output (MIMO) deep learning architecture with entity embeddings to predict delivery times across five journey stages. The system reduced training time to one-fifth of previous approaches, cut training memory by 50%, and improved order-to-assignment accuracy by 30%, enabling more efficient logistics operations across 500,000 restaurants and 1 million delivery partners.

Building on our earlier analysis of General Motors' AI deployment, this updated case study examines how the automotive giant developed a comprehensive AI-powered supply chain risk prediction platform over four years that prevented at least 75 factory stoppages in one year. The system combines digitized multi-tier supplier mapping, machine learning tracking, real-time news scanning of thousands of daily articles, and predictive modeling to provide days or weeks of advance warning for disruptions, including successfully predicting Hurricane Helene's impact on supplier Auria Solutions. GM's AI strategy extends beyond supply chain to autonomous vehicles, customer service, and manufacturing, supported by partnerships with Microsoft and Boston Consulting Group.

Toyota partnered with IBM Watson Advertising to launch the automotive industry's first cognitive ads for the Prius Prime plug-in hybrid. Using Watson's natural language processing on The Weather Channel app and weather.com, the campaign enabled one-to-one conversations with prospective buyers, achieving 3x Google Rich Media interaction time benchmarks, 6,000 user conversations, and a 20% increase in purchase consideration among the target demographic.

Chatmantics deployed IBM Voice Agent with Watson to automate contact center operations, handling over 250,000 inbound calls monthly. The platform reduced client acquisition costs by 60-70% and achieved production deployment in three months with lean operational teams.

Citigroup deployed agentic AI capabilities in its proprietary Citi Stylus Workspaces platform, piloting with 5,000 employees across the US, UK, India, and Canada in September 2025. The platform, powered by Google Gemini and Anthropic Claude models, enables employees to complete complex, multi-step tasks—such as client research, profile building, and translation—with a single prompt, eliminating the need for separate human direction at each stage. The initiative represents a shift from AI as a support tool to autonomous agents capable of executing workflows independently.

Copel, a Brazilian energy utility, deployed an AI agent powered by Gemini Pro 1.5 to enable natural language queries against their on-premises SAP ERP system. The solution reduced query processing time from one day to near real-time, eliminating manual data wrangling and empowering employees to self-serve financial and operational insights.

Siemens Energy deployed the Mendix low-code platform with integrated AI capabilities across finance, manufacturing, and operations, delivering over 200 applications that replaced legacy systems and reduced manual work. The deployment scaled from 4,000 to 40,000 users and generated millions of euros in added value, with generative AI applications like Chat UI reaching 30,000 concurrent users.

The Chinese University of Hong Kong deployed DCAN (Deep Contour-Aware Networks), a deep convolutional neural network for automated gland segmentation in colorectal cancer histology images. The system won the 2015 MICCAI Gland Segmentation Challenge, achieving an F1 score of 0.912 on benign tissue and 0.716 on malignant tissue, while training 100 times faster on NVIDIA TITAN X GPUs compared to CPUs.

Georgia-Pacific deployed ChatGP, a generative AI chatbot built on Amazon Bedrock and Claude, to address critical knowledge loss as experienced operators retire. The system integrates real-time IoT data from 90,000 sensors with institutional knowledge captured from veteran workers, providing operators instant troubleshooting guidance and reducing downtime across 140 manufacturing facilities.

Block deployed codename goose, an open-source AI agent powered by Claude 3.5 Sonnet on Databricks, saving 75% of engineers 8-10+ hours weekly and enabling 4,000 of 10,000 employees to independently solve data problems. The deployment contributed to a strategic workforce reduction from 10,000 to 6,000 employees while doubling gross profit per employee from $500k to $1M, with projections to reach $2M by 2026.