On July 18th, DEEPEXI TECH (01384) initiated the first public beta testing phase for its DeepWorks Enterprise AI Agent Platform. This launch was accompanied by a significant upgrade to its core product suite, culminating in the formation of the DeepexiOS AI-grade enterprise operating system. This system integrates the DeepWorks Enterprise AI Agent Platform, the Deepexi Enterprise Large Language Model, and the Deepology enterprise ontology data and Skills collection.
The DeepWorks Enterprise AI Agent Platform represents a new generation of enterprise-level AI agent solutions. It builds upon the foundation of DEEPEXI TECH's FastAGI platform, which has already served over 400 major enterprise clients. The platform has undergone extensive technological enhancements and capability upgrades. It incorporates a Harness architecture, improves Agent Team collaboration capabilities, adds a complex task Loop mechanism, integrates with enterprise organizational modules, and offers Token-based productivity cloud services.
The DeepWorks platform supports integration with various general-purpose foundational models, providing common office automation capabilities. When further connected with DEEPEXI TECH's proprietary Deepexi Enterprise Large Language Model, it enables continuous governance and iteration of the enterprise's knowledge semantic base. This allows for precise planning of long-term tasks based on specific enterprise knowledge, creating a complete capability chain from requirement generation and task planning to code engineering within the enterprise's unique semantic context.
The Deepexi Enterprise Large Language Model has a total parameter scale of 400 billion, with 23 billion activated parameters. It can be accessed via Token-based cloud services through the DeepWorks platform. For clients with high confidentiality requirements, the company also offers a version with quantized, smaller parameters suitable for deployment on private enterprise or even desktop-level computing units.
The Deepexi model focuses on an ontology-based paradigm, possessing the autonomous capability for enterprise data governance. It can iteratively refine the enterprise knowledge semantic base, enabling AI to more reliably understand enterprise objects, business rules, process relationships, and industry semantics. Simultaneously, it continuously compiles enterprise data foundations and daily operational data into an organizational-level semantic base, forming a "living memory" semantic foundation essential for practical AI implementation. This drives the genuine integration of AI into enterprise production and operational decision-making processes.
Deepology enterprise ontology data and the Skills collection comprise over 2,000 pre-built Skills, developed from DEEPEXI TECH's extensive experience serving numerous clients. These cover business domains across manufacturing, retail, scientific research, and public services. The Deepology collection already includes sets like DeepSense for manufacturing, DataSense for operational decision-making, and FDESense for IT infrastructure and data governance services.
Enhancing Enterprise AI Implementation Capabilities
This product suite upgrade significantly strengthens DEEPEXI TECH's capabilities for enterprise AI implementation. It facilitates the evolution of enterprise AI from merely utilizing general-purpose capabilities to achieving deep coupling with business semantics, and from isolated tool applications to a governable, sustainably evolving organizational-level productivity system.
Looking ahead, DEEPEXI TECH plans to continue deepening the integration of its product capabilities with specific industry scenarios, aiming to assist more enterprises in achieving productivity upgrades.
Comments