Samsung Electro-Mechanics Pioneers AI Collaboration with Agent-to-Agent System
Samsung Electro-Mechanics (009150.KS) is embarking on an ambitious endeavor to forge a collaborative framework wherein diverse artificial intelligence (AI) agents exchange tasks autonomously.
This innovative “agent-to-agent (A2A)” paradigm is set to transform operations by enabling heterogeneous AI agents to interact dynamically, thereby enhancing automation efficiencies.
This initiative comes at a time when the company is experiencing a surge in order volumes amidst the burgeoning AI landscape.
Information technology sources reported on the 20th that Samsung Electro-Mechanics is formulating a comprehensive strategy to integrate A2A by developing its proprietary AI agent alongside SAP’s AI agent “Joule.”
The in-house AI will be geared toward functionalities such as travel settlements, expense management, and facilitating document exchanges with suppliers.
The completion of this proprietary AI agent is slated for October, utilizing Samjong KPMG’s advanced AI development platform known as “Kaizen.”
Joule, an AI agent provided by SAP, operates in conjunction with SAP’s enterprise resource planning (ERP) system.
Following the second quarter of next year, Samsung Electro-Mechanics plans to embrace the SAP Business Technology Platform (BTP), thereby implementing an IT ecosystem that leverages Joule across various ERP functionalities.
Currently, Joule’s application is confined to the division responsible for ERP-related developments, but plans for a significant expansion of its utilization are underway for the coming year.
The company anticipates achieving a seamless integration of workflow automation with both AI agents once the proprietary system is finalized this year and Joule’s operational capabilities are broadened post-second quarter of next year.
Joule is expected to facilitate tasks intimately connected to product manufacturing. Samsung Electro-Mechanics has already incorporated supply chain management (SCM) and production planning data into its ERP framework, allowing Joule to contribute to analytical tasks tied to raw material transactions and product shipments.
Simultaneously, the proprietary AI agent will address functionalities outside the ERP scope, such as management tasks handled by Bizplay.
While Joule will focus on automating production-related activities, the proprietary AI agent will bolster management support functions, aiming to mitigate human intervention throughout operational processes.
This collaboration between the two AI agents is designed to allow personnel to concentrate on strategic decision-making with enhanced AI assistance.
In essence, A2A refers to the interactivity among AI agents from various platforms, enabling them to execute tasks either horizontally or vertically.
To simplify, A2A embodies the protocol through which OpenAI’s ChatGPT and Anthropic’s Claude delegate responsibilities, collaborate on resources, and perform tasks on behalf of humans to achieve collective objectives.
A critical aspect of this initiative is the necessity for effective communication among AI agents. The concept of A2A gained substantial traction following Google’s introduction of an open A2A communication standard in April of last year.
AI, previously limited to accessing pre-existing knowledge and publicly available data, experienced a paradigm shift with the advent of Anthropic’s Model Context Protocol (MCP) in 2024, which standardized communication between AI and computer operating systems, software, and files.
This development has effectively created an environment in which AI can traverse all layers of a computer for task execution.
Taking it further, Google unveiled the A2A protocol, enabling inter-agent communication. Historically, AI systems developed on disparate frameworks and programming languages require extensive coding to facilitate interaction; a challenge that escalates with the addition of more AI agents.
In contrast, Google’s A2A protocol employs a unified communication standard atop web standards, thereby establishing a common language for AI agents to confirm and exchange tasks efficiently.
A2A is garnering interest as a viable strategy for enterprises aiming to elevate the efficacy of AI deployment on a corporate scale.
Given that each AI agent possesses distinct strengths, the capacity for superior automation outcomes varies by industry and specific tasks.
Thus, a high benchmark score for a global tech conglomerate’s AI does not necessarily equate to improvements in organizational performance. Consequently, businesses are eager to deploy specialized AI agents in appropriate areas.
The flexibility inherent in A2A also promises cost-effectiveness. By utilizing both high-performance and lower-tier AI models in tandem based on task significance, companies can avert excessive expenditures on tokens.
Moreover, the compatibility between AI agents facilitates expedited development timelines, as there is no need for incremental deployment of function-specialized AI across existing IT infrastructures.
Samsung Electro-Mechanics’ methodical development of AI agents for both SAP ERP and non-ERP environments exemplifies this approach. A chief executive from an AI startup remarked, “Companies are inclined towards A2A for linking critical software like SAP or Salesforce with external AI agents.”
Such advancements in efficiency and cost management are pivotal in the rationale behind the adoption of AI agents by corporations.
A survey conducted last year by McKinsey & Company revealed that enhancing efficiency and revenue growth are the primary motivations for implementing AI solutions internally.
Among the 109 responding firms, a remarkable 84% highlighted improved efficiency, while 82% anticipated revenue growth as key benefits of AI integration.
In conclusion, as observed in the case of Samsung Electro-Mechanics, various domestic enterprises are rapidly exploring avenues for A2A implementation.
In response to this burgeoning corporate demand, the system integration (SI) sector is initiating efforts within the A2A domain. GS Neotek, for instance, recently showcased its AI agent platform “AIdea” demonstrating A2A technology to a distribution company.
Discussions regarding a proof-of-concept contract for A2A adoption are ongoing for the third quarter of this year.
An official from a SI firm noted, “Inquiries about establishing A2A have started filtering in since late last year,” adding, “The pace has picked up, with approximately one inquiry per quarter, suggesting that firms are beginning to express genuine interest.”
Source link: En.sedaily.com.
