FeasiGenAI

Inspiration Organizations struggle to manually analyze large volumes of BRDs to identify automation opportunities. This process is time-consuming, SME-dependent, and inconsistent. We wanted to eliminate that bottleneck by building an AI agent that reads a BRD like an experienced automation consultant instantly and at scale. What It Does FeasiGenAI reads a BRD/PDD document and delivers a full automation feasibility report. It identifies the process type, detects automation opportunities with a confidence score, flags process gaps, recommends the best automation tool (UiPath, Power Automate, API, Low code), and suggests the right coding language (Python, C#, VB.NET). The output is a structured recommendation report delivered automatically. Built on UiPath Maestro + BPMN, FeasiGenAI uses four specialized AI agents

Challenges We Ran Into BRDs vary wildly in format — narratives, tables, flowcharts — making consistent extraction difficult. Defining a meaningful confidence score required careful criteria design. Standardizing data handoffs between agents in Maestro and deciding when to trigger human review for borderline cases were the key technical and design challenges we navigated. Accomplishments We're Proud Of We built a fully agentic, end-to-end BRD analysis pipeline with zero manual SME involvement. The modular agent design allows independent updates without breaking the workflow. Most importantly, FeasiGenAI doesn't just check feasibility — it actively finds process gaps, making it more valuable than a simple yes/no automation checker. What We Learned Single-responsibility agents produce far more reliable and explainable outputs than broad, multi-purpose ones. Document quality directly impacts recommendation quality — which itself is a useful signal. Stakeholders trust recommendations more when confidence scores come with clear reasoning rather than just a number. What's Next for FeasiGenAI We plan to add a dedicated "Detect Process Gaps" agent, a structured output report template, and a weighted confidence scoring rubric. Future versions will support multi-document ingestion, SharePoint/Confluence integration, an ROI Estimation agent, and domain-specific variants for BFSI, Insurance, and Healthcare.