GenHAPI: Conversational Orchestration over MCP to Unlock Complex Legacy Workflows

Legacy enterprise applications remain functionally powerful but are often difficult to use, relying on fragmented workflows, dense forms, and expert knowledge. GenHAPI introduces a chat-to-action orchestration layer that enables users to interact with legacy systems through natural-language conversation while preserving control, auditability, and compliance. The system translates conversational requests into structured tasks defined in a declarative intent knowledge base and executes them via the Model Context Protocol (MCP) under explicit human confirmation. GenHAPI combines conversational orchestration, parameter governance, and optional vision-based UI grounding to support mixed-initiative interaction across heterogeneous enterprise environments. We report the system architecture, an intent engineering pipeline derived from real user traces, and a multi-agent workflow for safe tool invocation. Preliminary evaluation results include system-level benchmarks and a pilot user study, indicating reduced interaction complexity and improved user control compared to traditional workflows.