CAMBRIDGE, MA, May 5, 2026 – JuliaHub launched Dyad 3.0 and raised $65 million series B funding to expand AI-based engineering tools for physical systems. The software brings autonomous AI agents into digital design, simulation, testing and embedded code generation for industrial machines, including heat pumps, satellites and semiconductors. JuliaHub said several Fortune 100 companies use Dyad and Julia across aerospace, government, automotive, HVAC and utilities.
Daniel Freeman, who led the series B round for Dorilton Capital, commented: “Systems modeling is one of the most strategically important layers of the AI-native engineering stack, because it is where physics, control logic, and AI converge. JuliaHub has built something extraordinary with Dyad: a platform that doesn’t just model systems, but compiles them, taking engineers from concept to production control code in a single environment. We believe JuliaHub has the potential to become one of the defining companies in Physical AI, and we’re proud to back the team as they accelerate Dyad’s path to market.”
Hardware Engineering and AI
JuliaHub said physical engineering has not adopted AI-based workflows at the same rate as software development. The company cited Claude Code, Codex, and Gemini as examples of AI tools used in software development. It said industrial engineers still face limits from older modeling and design tools. JuliaHub also cited a McKinsey estimate that $106 trillion in cumulative investment will be needed through 2040 for new and updated infrastructure.
Dyad gives engineering teams an AI-based environment to model, test, and validate industrial systems. Dyad 3.0 builds on Dyad 1.0, launched in June 2025, and Dyad 2.0, launched in December 2025. The software connects autonomous agents with scalable physics simulations, controls, safety analysis, and embedded code generation. Dyad can support digital twin development, controller tuning for deployment scenarios, and hardware design iteration for systems such as wastewater facilities and automobiles.
“It’s not about helping engineers complete one small task at a time. It’s agentic engineering at scale, where teams can feed a full specification to Dyad and have it design the complete system. Spec in. Design out,” said Viral Shah, CEO of JuliaHub.
Digital Twins and Scientific Machine Learning
Dyad’s cloud-based agents are designed to scan scientific information and improve models over time. JuliaHub said AI-automated lab testing can help align models with physical systems. Streaming data and Scientific Machine Learning (SciML) allow models to update as systems collect real-world data. Dyad’s simulation ecosystem and language return results to engineers for process checks, assumption reviews, customer requirement checks, and safety review.
Prith Banerjee, senior vice president of innovation at Synopsys commenting on the partnership with JuliaHub said, “Dyad is transforming system-level engineering by combining scientific AI, agentic modeling, and a powerful compilation pipeline into a unified workflow. Integrated with Synopsys simulation software Ansys TwinAI, it enables high fidelity hybrid digital twins by integrating physics-based simulation with data-driven models. What once required extensive manual effort can now be done far more efficiently, accelerating the entire digital engineering lifecycle and redefining how intelligent, software-defined systems are designed and validated.”
AI for Physical Systems
General-purpose AI cannot ensure that a model follows the laws of physics. Modeling errors in physical engineering can affect bridges, batteries, and other safety-critical systems. In recent agentic benchmarking for chemical process modeling, JuliaHub said general LLM systems such as Codex, Claude Code Opus, and Gemini completed little beyond the initial setup. Dyad automated most of the process for creating model-predictive controllers to optimize chemical plant yields, a task that takes weeks.
“There is a disruptive transition occurring in engineering system design software, and Dyad is on the cutting edge. Previous generations of tools do not provide the promised productivity or integration to unlock the value of AI. With Dyad, you can model the physics, develop controls algorithms with auto code generation, and create accurate digital twins and surrogates for rapid development of deep learning inference models, all enabled by AI. Dyad operates where physics meets analytics, and customers and shareholders win!” said David Joyce, former CEO of GE Aviation and Vice Chair of GE.
Dyad’s modeling language is designed for AI agents to process. JuliaHub said its logic is grounded in physics, allowing agents to reason about fluid movement, wind speed, temperature, gravity, and other forces. This approach produces physically valid models for engineering review. In a partnership with Binnies and Williams Grand Prix Technologies, JuliaHub developed a SciML-powered digital twin for water distribution systems. The system uses four sensor inputs to predict pump faults with more than 90% accuracy.
“Dyad represents a step-change for the water industry, enabling a move from reactive operations to predictive, system-level decision making,” said Tom Ray, director of digital products & services (Digital Twins & AI) at Binnes. “It has the potential to transform how companies model real-world complexity, predict failure, and optimize performance every day.”
Launch Event
Dyad 3.0 will be unveiled at a live event on May 19. The event will include product demonstrations and customer experiences across aerospace, HVAC, utilities, and robotics.
Source: JuliaHub
About JuliaHub
JuliaHub, founded in 2015 and headquartered in Cambridge, MA, provides cloud software and tools for scientific and technical computing. The company supports pharmaceutical, aerospace, automotive, electronics and manufacturing customers that use modeling, simulation and computational analysis in their research and engineering work. JuliaHub develops and maintains a platform for building and running applications in the Julia programming language and offers tools for Scientific Machine Learning, digital-twin modeling, circuit simulation and drug-development workflows. The company reports more than 10,000 global users and employs about 110 people worldwide. Its stated mission is to support organizations addressing scientific and engineering problems by delivering secure computing environments and mathematical and machine-learning capabilities.