AgentX: Towards Orchestrating Robust Agentic Workflow Patterns with FaaS-hosted MCP Services

2025-09-09

AI Summary

Agentic AI systems suffer from insufficient robustness, hallucination, and difficulties in maintaining contextual consistency—particularly when coordinating multiple tools, planning long-horizon tasks, and preserving state across extended interactions. To address these challenges, this paper introduces AgentX, a novel agent workflow paradigm featuring a three-tier collaborative architecture: “Phase Design → Stepwise Planning → Precise Execution,” integrating dedicated Phase Designer, Planner, and Executor roles. Furthermore, we propose two lightweight, highly elastic deployment strategies for Model Context Protocol (MCP) services—leveraging both MCP-native interfaces and Function-as-a-Service (FaaS) cloud functions. Extensive experiments across three real-world application scenarios demonstrate that AgentX significantly improves task success rates while reducing end-to-end latency and API invocation costs. It outperforms representative baselines—including ReAct and Magentic One—establishing a scalable, reproducible, and production-ready engineering framework for complex agentic AI systems.

Technology Category

• Multiagent Systems: Agent Communication
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Generative Artificial Intelligence (GenAI) has rapidly transformed various fields including code generation, text summarization, image generation and so on. Agentic AI is a recent evolution that further advances this by coupling the decision making and generative capabilities of LLMs with actions that can be performed using tools. While seemingly powerful, Agentic systems often struggle when faced with numerous tools, complex multi-step tasks,and long-context management to track history and avoid hallucinations. Workflow patterns such as Chain-of-Thought (CoT) and ReAct help address this. Here, we define a novel agentic workflow pattern, AgentX, composed of stage designer, planner, and executor agents that is competitive or better than the state-of-the-art agentic patterns. We also leverage Model Context Protocol (MCP) tools, and propose two alternative approaches for deploying MCP servers as cloud Functions as a Service (FaaS). We empirically evaluate the success rate, latency and cost for AgentX and two contemporary agentic patterns, ReAct and Magentic One, using these the FaaS and local MCP server alternatives for three practical applications. This highlights the opportunities and challenges of designing and deploying agentic workflows.

Problem

• Addressing complex multi-step tasks in Agentic AI systems
• Managing long-context history tracking to prevent hallucinations
• Orchestrating numerous tools effectively in agentic workflows

Innovation

• AgentX workflow with stage designer, planner, executor
• MCP tools deployed as cloud FaaS services
• Competitive success rate, latency, cost versus ReAct