Supplier searches often begin too narrowly, lacking a structured, evidence-based approach. This workflow utilizes an AI agent with three specialized tools: a UN Comtrade scraper for macro trade signals, an IndiaMART scraper for supplier discovery, and a GST verification tool for registration checks. The process starts by assessing the overall trade volume for a product category using official trade data. If the category shows significant trade, the agent then searches marketplaces for potential suppliers in a specified region. Each supplier candidate's registration number is then validated through a public index. This staged approach ensures that resources are only spent on promising leads, and the system can gracefully handle incomplete or invalid data. The AI agent is given a clear decision policy, guiding its actions without granting unchecked authority. For instance, it confirms the user's request parameters before proceeding and treats trade data as context, not definitive proof of a company's legitimacy. If the macro trade signal is weak, the workflow stops early. When searching for suppliers, the agent distinguishes between individual product listings and unique companies, using GSTIN as a primary deduplication key. Crucially, all verification states, like "verified," "not_indexed," or "invalid_format," are preserved, allowing for nuanced assessment rather than a simple pass/fail. The final report presents raw evidence, clearly categorizing candidates for further human review or manual verification. This workflow prioritizes transparency, allowing users to understand the basis of each decision made by the agent.