Selected work / Production application
Reducing the cost of customer-brief preparation with an AI assistant
Enterprise Microsoft Teams assistant using Azure OpenAI, Semantic Kernel, .NET, and existing business data and services.
Reported before-and-after preparation times for the specific recurring workflow. These figures are not a controlled study or a claim about all users and tasks.
Problem
Preparing customer briefs required hours of manually gathering information and preparing a brief from material spread across existing enterprise systems. The goal was to shorten the preparation task using tools employees already had.
Constraints and decision
The assistant had to work through Microsoft Teams and existing services, while respecting existing permissions and integration requirements. Azure OpenAI and Semantic Kernel supplied language-model capability; .NET integrations connected it to business information. The application handled this preparation task; it was not designed as a general-purpose agent.
Contribution and outcome
Designed and built the enterprise assistant and its integrations as part of a production engagement. The reported workflow improved from more than three hours to fifteen minutes or less. Delivery included production use and a phased rollout, not just a standalone prototype.
Evidence and limits
The metrics are drawn from the reported professional outcome and should be read as workflow-specific. Employer or customer-sensitive code, data, and internal operational records are not published here. The reported result is described here, but the private implementation and operational records cannot be independently reviewed.