中国·新威尼斯v0008总站(股份)有限公司-Official website

SemiMind

SemiMind is a large language model platform in the semiconductor industry driven by Knowledge Base & AI Agents. It leverages models like DeepSeek-R1 to build an open, flexible, and scalable intelligent R&D ecosystem. SemiMind extends AI applications in areas such as chip design automation, yield enhancement, and intelligent software platforms. It empowers the semiconductor industry with autonomous thinking capabilities, reconstructs R&D processes intelligently, breaks down data silos, and drives the industry towards more efficient, cost-effective, and intelligent development.

undefined

Advantages

Multimodal Reasoning Capability:: Integrate text, data, and charts for multidimensional analysis, delivering highly reliable conclusions.

Unleash Innovative Potential:: Free engineers from repetitive tasks to focus on high-value innovation.

Accurate Intent Understanding:: Natural language interaction aligns with engineers' thinking, responding to complex needs in seconds.

Lower Technical Barriers:: Enterprises can quickly adopt top-tier company experiences to accelerate catch-up.

Drive Industry Synergy and Innovation:: Break down data silos across chip design, manufacturing, and testing stages to facilitate knowledge sharing and collaborative optimization.

Key Features

Knowledge Retention and Reuse with Query-Based Insights

Integrate industry know-how and vast process data to build a specialized knowledge base that breaks through experiential barriers. By storing semiconductor manufacturing data such as CP, FT, and WAT, it allows for rapid correlation of yield issues with each process steps and recommends solutions from similar cases.

Support for Custom AI Agent Construction with Local Inference

By leveraging a low-code or no-code platform, users can quickly build customized functional modules and AI agents for specific expertise, such as test plan generation, creation of equipment O&M documentation, real-time anomaly detection for process parameters, multi-source data analysis, and root cause identification, This enables agile responses to requirements.

Optimization of Engineering Processes to Build Construction of Customer-Specific Analytical Methodologies

The upgrade of EDA software with AI capabilities allows for flexible integration with other platforms, providing personalized recommendations, automated process management, and real-time data analysis. This helps users complete tasks more effectively and boosts work efficiency.

Application Scenarios

Test Plan Generation

Utilize large models combined with RAG technology to swiftly analyze standard documents and automatically generate test plans that align with device characteristics and standards.

Enhance operator efficiency and reduce error risks to ensure the standardization and consistency of testing processes.

AI-Driven Software Solutions

INF-AI Integration:
support for expressing requirements in natural language; seamless,
end-to- end workflow from data to model to service; automated labeling of common
defect classes; industry-leading accuracy in small-object and novel-class recognition; fully traceable and controllable service deployment.

DEG Integration:
perform complex tasks like chart generation and statistical analysis through
simple natural language interaction; offer built-in data governance engine; leverage FAB historical data to deliver instant, experience-based insights; breaking down traditional barriers to knowledge sharing.

iCASE Integration:
accumulate expert experience through a full-lifecycle data asset system; build an Agentic AI–driven
hub for defect analysis; realize second-level traceability and root-cause
identification of semiconductor manufacturing defects and provide yield optimization recommendations.