Amnon Shashua's doubleAI Unveils Maxwell Code Optimization Engine
Prof. Amnon Shashua's startup doubleAI unveiled Maxwell, an advanced AI processing optimization engine designed to lower computing costs by modifying live code across hardware layers.
Amnon Shashua's new AI startup, doubleAI, has unveiled its advanced code optimization engine known as Maxwell, launching it in early access for developers. Unlike traditional software development tools, Maxwell does not compete with products like Anthropic's Claude Code or OpenAI's Codex. Instead, it serves as an AI processing optimization system designed to significantly lower costly computing expenses for enterprises.
Optimizing Hardware and Middleware Layers
The key differentiator of doubleAI's new engine is its ability to modify live code across hardware and middleware layers while software runs. While other language models translate developer commands into source code, Maxwell operates differently. Software and hardware activity, along with moving data, directly generate machine-language instructions that refine over time, performing runtime adjustments previously handled by specialized software components.
"Artificial intelligence has changed the way software is written, but not the way it operates," said Prof. Amnon Shashua, co-founder of doubleAI.
According to the startup, this approach offloads considerable processing activity from chips, saving valuable computing tokens and expensive graphic processing units (GPUs). Shashua noted that the system operates seamlessly across both AI-focused GPUs and central processing units (CPUs) handling storage and application management.
Mathematical Verification and Efficiency
The development engine includes a specialized component utilizing non-LLM algorithms that acts as a verifier, checking written code and confirming correctness with mathematical certainty. This verification mechanism builds upon doubleAI's recent breakthrough in solving the Longest Path Problem, a major unsolved challenge in graph theory, achieving this feat at a remarkably low computational cost under $2,000.
Backed by $200 million in funding from prominent venture funds and a valuation reaching $1 billion, doubleAI continues to push technological boundaries. Investors like Dell Technologies Capital highlight the startup's unique approach to computational efficiency, positioning it at the forefront of advanced artificial intelligence innovation.
