SSE set out to improve the way it supports customers, particularly around billing, urgent enquiries, and complex business accounts in a tightly regulated energy market. Traditional, rule-based chatbots weren’t meeting expectations because: - Customer queries are often nuanced and don’t follow simple decision trees. - Business customers need detailed, sometimes bespoke information that spans multiple systems. - Customers phrase questions in many different ways and don’t use SSE’s internal terminology. At the same time, SSE must comply with strict regulatory requirements and protect all customer data. That means answers must be accurate, compliant, and traceable—there’s very little tolerance for the AI “making up” responses. To address this, SSE used Microsoft Copilot Studio and Azure OpenAI to build “Nero,” a conversational virtual assistant that: - Understands natural language and multi-part questions. - Pulls information from multiple content sources via Azure AI Foundry and Retrieval Augmented Generation (RAG). - Operates within clearly defined guardrails to keep responses compliant, accurate, and secure. This combination allows SSE to deliver a smoother, more intuitive customer experience while still meeting regulatory expectations.

Nero is embedded directly into SSE’s website and is available 24/7 to help customers manage their energy accounts. In practice, customers can use Nero to: - View bills and outstanding balances. - Make payments and query specific charges. - Check usage metrics and compare consumption month to month. - Sign up for alerts, such as notifications when usage is unusually high. Under the hood, Nero uses Microsoft Copilot Studio for natural language understanding and Azure OpenAI (via Azure AI Foundry) to generate clear, contextual answers. Retrieval Augmented Generation (RAG) ensures Nero only responds based on SSE’s approved knowledge sources, which helps maintain accuracy and regulatory compliance. The impact so far includes: - Around 280 customer conversations handled per day. - A 53% increase in positive customer reactions within the virtual assistant. - A 99% improvement in intent recognition and answering, effectively eliminating missed messages. By handling routine and moderately complex queries, Nero frees human agents to focus on higher-value, more complex cases. Customers increasingly choose to use the assistant because it’s fast, available at any time, and aligned with how they naturally ask questions.

SSE chose Microsoft Copilot Studio and Azure OpenAI for a combination of technical, regulatory, and strategic reasons: - **Compliance and security:** Azure services are built on enterprise-grade infrastructure and aligned with global data regulations. SSE also worked with Microsoft Unified to implement a multi-layered security model that protects customer accounts and information. - **Responsible AI:** Microsoft’s Responsible AI Standard helps SSE ensure Nero’s responses are inclusive, reduce bias, and stay within defined guardrails. - **Native integration:** Microsoft’s AI capabilities are integrated across the Azure platform and Microsoft 365, making it easier for SSE to adopt AI consistently across the business. - **Speed of innovation:** New AI models become available on Azure quickly after announcement, allowing SSE to experiment and deploy improvements without long delays. Beyond Nero, SSE is adopting an AI-first mindset: - Internally, Microsoft 365 Copilot has up to 90% daily usage among eligible staff, supporting tasks like coding, requirements analysis, and test case creation. - SSE is building an AI Centre of Excellence with Microsoft to share best practices and ensure responsible deployment across use cases. Looking ahead, SSE plans to evolve Nero into an omnichannel, multimodal assistant that can: - Support voice interactions. - Understand scanned documents and meter photos. - Provide more personalised, agentic analytics to help customers manage usage, reduce costs, and move closer to net zero. In short, SSE views Microsoft AI as a foundation for reimagining both customer interactions and internal operations, with a focus on practical, customer-centric outcomes rather than one-off experiments.