Case Studies

TP ICAP, a global financial markets infrastructure provider, deployed a generative AI compliance solution on AWS in just four weeks, reducing regulatory document review time from approximately one month to minutes. The system, built using Amazon Bedrock and Amazon Textract, enables non-technical compliance staff to upload documents and receive instant assessments, transforming a manual process that previously consumed weeks of analyst time.

Smartly, an advertising technology platform managing $5 billion in ad spend for 700+ brands, deployed self-hosted Llama 3.1 8B on Kubernetes to automate customer support ticketing and resolution messaging. The implementation reduced ticket creation time by 80%, cut resolution drafting time by 50%, and increased customer satisfaction scores while maintaining complete data control through on-premises deployment.

Swindon Borough Council deployed an AI-powered fly-tipping reporting system using AWS Rekognition and US Army object recognition data, reducing average cleanup time from 11 days to 4 days while saving over £28,000 annually in fuel and staffing costs. The system handles 300 monthly cases with 98% resident satisfaction, operating at £150/month.

The Arizona Department of Child Safety deployed a suite of GenAI tools—branded as SOLVE AI—to automate allegation classification, policy search, document processing, and DevOps workflows. Working with IBM Consulting and Microsoft Azure, the department reduced policy record removal time by 58%, cut call documentation burden by 2,800 hours annually, and achieved a 40% productivity gain across caseworkers.

Stone Ridge Technology deployed ECHELON 2.0, the first commercial reservoir simulator built from inception for NVIDIA GPUs, achieving 10-100x speedups over CPU-based simulators. In partnership with Eni, the system simulates an 11.2-million-cell reservoir model in 30 minutes on 8 NVIDIA A100 GPUs—a task that previously took days or weeks on traditional multi-core CPU clusters.

Swindon Borough Council deployed an AI-powered fly-tipping detection system using AWS Rekognition to analyze citizen-submitted photos, automatically classify waste types, estimate volumes, and optimize cleanup routes. The system reduced average cleanup time from 10 days to 4 days, saved £28,000 annually in fuel and staffing costs, and achieved 98% citizen satisfaction—all while running on AWS infrastructure costing just $230 per month.

Bonsai Robotics deployed vision-based AI autonomy across 50+ machines in U.S. and Australian nut orchards, doubling harvest speed while cutting operating costs by 45%. The system navigates dust, darkness, and GPS-denied environments using computer vision trained on 500,000+ acres, enabling 24/7 autonomous operation of shakers, sweepers, and tractors.

Filevine, a Salt Lake City legal tech company, now generates more quarterly revenue from AI products than its traditional SaaS platform—a rare milestone in enterprise software. The company's AI tools for document analysis, medical chronologies, and real-time deposition assistance drove 130% YoY AI revenue growth, fueling a $400M raise at $3B valuation. With 6,000 customers and 125,000 daily users, Filevine demonstrates how embedded AI in a system of record can outpace standalone SaaS.

Robinhood deployed FinCrimes Agent, a generative AI solution built on Amazon Bedrock, to transform financial crimes investigations. The multilayered system uses Anthropic's Claude Haiku and Sonnet models alongside DeepSeek to synthesize customer and transactional data into concise investigative summaries. The deployment achieved a 20% cumulative efficiency gain in investigative workflows, allowing analysts to identify illicit activity with greater focus and accuracy while maintaining rigorous standards for explainability and regulatory compliance.

Cisco deployed AgenticOps, an AI-driven framework for autonomous network operations, across its global enterprise networking portfolio. The system uses a proprietary Deep Network Model trained on 40 years of networking expertise to enable proactive troubleshooting, autonomous optimization, and sub-second mean-time-to-resolution across campus, branch, industrial, and data center networks.

Cisco deployed AgenticOps, an agent-first IT operating model that uses autonomous AI agents to manage network operations at machine speed while maintaining human oversight. Powered by the Deep Network Model—a domain-specific LLM trained on 40 years of Cisco expertise—and AI Canvas, a generative UI for cross-domain collaboration, AgenticOps enables IT teams to shift from reactive firefighting to proactive optimization across campus, branch, data center, and industrial networks. The platform integrates telemetry from Cisco Networking, Security Cloud Control, Nexus One, Splunk, ThousandEyes, and Secure Firewall to deliver autonomous troubleshooting, agentic workflow creation, and real-time insights.

Orange Group deployed Live Intelligence, a sovereign multi-LLM generative AI platform, to 90,000 employees across 26 operating companies. The telco built 150+ specialized AI assistants, trained 35,000 employees, and projects €300 million in value generation for 2025 across customer experience, network operations, and operational efficiency.