PMO | What is AI Agent and How It Impacts Project Management

​An AI agent is an autonomous system that utilizes artificial intelligence to perform tasks, make decisions, and interact with its environment. These agents can learn, adapt, and react to inputs while working towards specific goals. Examples include chatbots, self-driving cars, personal assistants, and recommendation systems, which often improve over time through experience.

Recently, Microsoft announced its AI agent for project management called the Multi-Agent Runtime Service (MARS). This service claimed to contain more than 40 agents that can handle various project management tasks, such as work breakdown, market research, and content summary. In the reference section, we have included a radio link for you to watch the demo [1]. The demo was particularly impressive for several use cases:

• Generating Status Reports: This feature seems useful as it collects all completed tasks and summarizes them automatically. This could be beneficial.
• Consolidating Comments: During the collaboration, numerous comments came from various sources and were from different places. Consolidating these into a summary can provide valuable insights.
• Task Analysis and Focus: Analyzing tasks and summarizing what we should prioritize and focus on next is a good concept.​

The agent concept is promising, as specific tasks can be automated, reducing our reliance on humans for those jobs. However, we must be cautious about which AI agents we choose to adopt for project management for the following reasons:

• Cost: While AI agents can automate some tasks, we must consider the costs involved. For instance, creating meeting minutes from a WebEx recording can be expensive and may not be practical in project management settings. 
• People Development: Project management requires fundamental skills, and relying too heavily on AI can hinder skill development. For example, using an AI agent to generate meeting minutes may undermine a project manager's ability to grasp discussion points quickly. Should a project manager depend on an agent for this instead of honing their skills? Additionally, we see some agents can search for components to assign to bug tickets. If someone is unsure which component to use, they should already have received training rather than being left to navigate a complex landscape. Efficiency in organizations means that looking things up should not become a constant necessity.​

While there is excitement around AI's potential, it's essential to examine the limitations behind that excitement. Currently, AI agents cannot complete tasks as effectively as humans can. Therefore, while AI agents are valuable, careful consideration must be given to which agents we choose to develop and implement.

Reference 

1. ​​Project Manager Agent in Microsoft Planner Demonstration, January 2025
2. How to Build a Multi-Agent RAG System (MARS) with OpenAI Swarm, November 2024
3. Unleashing the power of agents in Microsoft Planner,  November 2024