Microsoft rolled out a new multi-agent artificial intelligence (AI) system dubbed “Magnetic-One” on Monday. The tech giant called it a high-performance system that can activate multiple AI agents to complete complex tasks across web browsers locally on the device. It relies on a new framework that allows the AI model to access multiple methods and capabilities to complete tasks such as booking a ticket, purchasing a product online, or editing a document stored on the device. It is worth noting that Microsoft’s Magic-One program is an open source project and can be accessed by researchers and developers.
Microsoft introduces the Magnet-One device
Generative AI has made a huge leap in machine intelligence and its ability to generate output across text, image, audio and video formats. However, although modern AI systems are great at retrieving information, they are still weak at reasoning, especially when it comes to solving problems and completing tasks.
For this reason, AI agents, which can be understood as mini-programs capable of performing an action, have become an important extension of large language models (LLMs). Microsoft’s Magnet-One device also works on the same principle, as detailed in A Research paper. The company describes it as a “high-performance general agent system” designed to complete complex, multi-step tasks such as software engineering, data analysis, scientific research, and web navigation.
Magnet-One has a multi-agent architecture, which means a single LLM can activate multiple agents to complete a task. For this reason, the AI system activates a master agent called a coordinator. He directs four other agents where each agent specializes in one task.
Single magnetic workflow
Image source: Microsoft
For example, if the system is asked to reserve a ticket for a movie, the coordinator can run a vision agent that can look at the screen and process the visual information. Another person may have knowledge of web browsers and can handle navigation. The third may be to break down the claim into actionable steps, and the fourth may be able to handle financial transactions. By dividing the task between several specialized agents, the accuracy and speed of completion are increased.
The Magnet-One AI system is open source and accessible on GitHub here. It is available to researchers and developers, and can also be used for commercial purposes under a custom license from Microsoft. Besides, Microsoft has also released AutoGenBench, a tool that evaluates the performance of AI agents. It comes with built-in repeat and isolation controls to accurately test parameters.
(Except for the headline, this story has not been edited by NDTV staff and is published from a press release)
