Submission by Rairo

We built Sozo Discovery, a forensic material intelligence system that turns a single image into a real-time supply chain audit. It deconstructs any imave of a physical object into its core Bill of Materials (BOM) and uses the Firecrawl Search API to perform a live global risk assessment across the web’s "dark data." ​What problem does it solve? Static knowledge is a liability in global trade. Discovery solves the grounding problem by bridging the gap between a physical asset and the high-entropy machinery of global production. It replaces LLM "hallucinations" about material costs with verified, real-time data on spot prices, export duties, and regional port congestion. ​How does it use ElevenLabs and Firecrawl? ​Firecrawl Search API: This is the system’s "Digital Sensor." Rather than scraping a pre-defined list of URLs, the system uses the Search API to dynamically discover the most relevant, high-signal sources for each specific material in the BOM. It allows the agent to navigate real-time economic shifts, like a sudden tariff in Zimbabwe or a price spike in China and it provides a clean, structured context for the reasoning engine. ​ElevenLabs: Provides the Sovereign Interrogation Layer. The results from the Firecrawl Search are injected as dynamic variables into a low-latency voice session. This transforms a search result into a Forensic Dialogue, allowing the user to orally interrogate the supply chain, explore chokepoints, and strategize risk mitigation through natural speech.