Tip Of AI
Discover what you can do with AI
Archimedes is an agentic engineering system from Inviscid AI that searches for improved physical designs by writing designs as code and testing them in physics simulators. Users provide an engineering goal and constraints while multiple AI agent teams generate alternatives run simulations share successful ideas and continue improving the strongest results. The platform supports design problems involving fluid dynamics thermal management mechanical systems and physical control. In one drone propeller study six teams used twelve cloud machines to evaluate 1251 designs in twelve hours. The selected propeller worked in both air and water and achieved about 90 percent of the performance of specialist propellers in each environment. Other reported results include 10.9 percent lower data centre electricity use with estimated annual savings of 227000 dollars 25.5 percent lower turbine blade loss 2.5 times better backflow blocking for a valve 27 percent more lift for a wing and 42 percent lower spill risk for a mug. Archimedes also performs finer simulation checks and reports when an apparent improvement fails under more realistic conditions. The system is in early access and supports custom engineering studies. Reynolds is open source under the MIT licence.
