Global industrial R&D is currently manual, slow, and highly fragmented. This "Innovation Blindspot" causes up to 40% of deep-tech and industrial projects to fail due to missed prior-art patent risks, unviable material constraints, or unoptimized supply chains identified too late in the cycle. The Solution Drishti Chakra is an autonomous agentic suite (consisting of Shodh [Research], Manthan [Synthesis], and Parikshan [Audit]) that crawls global scientific journals and patent databases to identify market gaps, generates structured engineering specifications, and audits intellectual property viability. For this hackathon, we demonstrated its capabilities on one of the most mathematically challenging physics topics: Autonomous Anti-Gravity Propulsion Systems (Electrogravitics & Superconductor-based). The AMD Advantage Drishti Chakra is custom-built for the AMD ROCm 6.x ecosystem. By leveraging the industry-leading 192GB HBM3 memory and 5.3 TB/s bandwidth of the AMD Instinct™ MI300X (via Fireworks AI), our model loads entire multi-patent corpuses and detailed engineering papers simultaneously without performance bottlenecks. Our "Deep-Reach" logic powered by Gemma-2-9b enables deep, recursive reasoning steps that ensure no critical engineering bottleneck or patent landmine is overlooked. Market Potential Targeting a fraction of the global $500B enterprise R&D, compliance, and strategic management consulting market, Drishti Chakra functions as a "Startup-in-a-Box" for deep-tech founders and industrial pioneers.
Category tags: