The Landscape of Emerging AI Agent Architectures for Reasoning, Planning, and Tool Calling: A Survey
2024/04/17 by Tula Masterman, Masterman, Tula, Sandi Besen +5 · 3 voices · 81 citations
Computer Science · #Artificial intelligence #Computer science #Data science #Geography #Multi-Agent Systems and Negotiation #cs.AI #cs.CL
paper · pdf · doi:10.48550/arxiv.2404.11584
published in arXiv (Cornell University) (Cornell University)
openalex publication_date 2024/04/17 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Abstract
This survey paper examines the recent advancements in AI agent implementations, with a focus on their ability to achieve complex goals that require enhanced reasoning, planning, and tool execution capabilities. The primary objectives of this work are to a) communicate the current capabilities and limitations of existing AI agent implementations, b) share insights gained from our observations of these systems in action, and c) suggest important considerations for future developments in AI agent design. We achieve this by providing overviews of single-agent and multi-agent architectures, identifying key patterns and divergences in design choices, and evaluating their overall impact on accomplishing a provided goal. Our contribution outlines key themes when selecting an agentic architecture, the impact of leadership on agent systems, agent communication styles, and key phases for planning, execution, and reflection that enable robust AI agent systems.
Cited by
- PAGE-RAG: Evidence-Grounded Adaptive Graph Retrieval for Long-Document Question Answering
- HiLSVA: Design and Evaluation of a Human-in-the-Loop Agentic System for Scientific Visualization
- SAAG: Structured Agent Assessment and Grounding
- AutoSurrogate: An LLM-driven multi-agent framework for autonomous construction of deep learning surrogate models in subsurface flow
- Distributional AGI Safety
- Magentic-UI: Towards Human-in-the-loop Agentic Systems
- Small Language Models are the Future of Agentic AI
- Addressable Recall Compaction for Long Context-Window Control in AI Agents
- Agent Team Work Zone: An Automated, Persistent Workspace for Long-Lived Claude Code Agent Teams
- Bridging the Gap on AI-Assisted Scientific Software Development Through Transparency and Traceability
- Explainable and Fine-Grained Safeguarding of LLM Multi-Agent Systems via Bi-Level Graph Anomaly Detection
- Toward Agentic Environments: GenAI and the Convergence of AI, Sustainability, and Human-Centric Spaces
- A Practical Guide for Designing, Developing, and Deploying Production-Grade Agentic AI Workflows
- CKG-LLM: LLM-Assisted Detection of Smart Contract Access Control Vulnerabilities Based on Knowledge Graphs
- GTM: Simulating the World of Tools for AI Agents
- Measuring Agents in Production
- An Empirical Study of Agent Developer Practices in AI Agent Frameworks
- SemAgent: Semantic-Driven Agentic AI Empowered Trajectory Prediction in Vehicular Networks
- Conversational No-code, Multi-agentic Disease Module Identification and Drug Repurposing Prediction with ChatDRex
- Multi-Agent Deep Research: Training Multi-Agent Systems with M-GRPO
- TAMAS: Benchmarking Adversarial Risks in Multi-Agent LLM Systems
- OceanAI: A Conversational Platform for Accurate, Transparent, Near-Real-Time Oceanographic Insights
- OracleAgent: A Multimodal Reasoning Agent for Oracle Bone Script Research
- Agentic AI: A Comprehensive Survey of Architectures, Applications, and Future Directions
- Agent Data Protocol: Unifying Datasets for Diverse, Effective Fine-tuning of LLM Agents
- Branch-and-Browse: Efficient and Controllable Web Exploration with Tree-Structured Reasoning and Action Memory
- A Survey of Data Agents: Emerging Paradigm or Overstated Hype?
- ToolDreamer: Instilling LLM Reasoning Into Tool Retrievers
- Empowering Real-World: A Survey on the Technology, Practice, and Evaluation of LLM-driven Industry Agents
- Natural Language Tools: A Natural Language Approach to Tool Calling In Large Language Agents
- AI for Service: Proactive Assistance with AI Glasses
- NetMCP: Network-Aware Model Context Protocol Platform for LLM Capability Extension
- From Craft to Constitution: A Governance-First Paradigm for Principled Agent Engineering
- Tool Calling for Arabic LLMs: Data Strategies and Instruction Tuning
- What makes prompts a graph: necessary and sufficient conditions for prompt graph engineering
- Scaling LLM-Driven Multi-Agent Systems: Design Principles and Architectural Scalability Analysis
- RoboSeek: You Need to Interact with Your Objects
- Generalizability of Large Language Model-Based Agents: A Comprehensive Survey
- AgenticIE: An Adaptive Agent for Information Extraction from Complex Regulatory Documents
- AgentX: Towards Orchestrating Robust Agentic Workflow Patterns with FaaS-hosted MCP Services
- TableMind: An Autonomous Programmatic Agent for Tool-Augmented Table Reasoning
- Finance-Grounded Optimization For Algorithmic Trading
- AgenTracer: Who Is Inducing Failure in the LLM Agentic Systems?
- Situating AI Agents in their World: Aspective Agentic AI for Dynamic Partially Observable Information Systems
- AI Compute Architecture and Evolution Trends
- Universal Deep Research: Bring Your Own Model and Strategy
- BetaWeb: Towards a Blockchain-enabled Trustworthy Agentic Web
- Towards Reliable Multi-Agent Systems for Marketing Applications via Reflection, Memory, and Planning
- BlindGuard: Safeguarding LLM-based Multi-Agent Systems under Unknown Attacks
- ToPolyAgent: AI agents for coarse-grained bead-spring topological polymer simulations
- A Survey on Agent Workflow -- Status and Future
- Blueprint First, Model Second: A Framework for Deterministic LLM Workflow
- Agentic AI framework for End-to-End Medical Data Inference
- Autonomous Computer Vision Development with Agentic AI
- Building crypto portfolios with agentic AI
- SEALGuard: Safeguarding the Multilingual Conversations in Southeast Asian Languages for LLM Software Systems
- A Survey on Autonomy-Induced Security Risks in Large Model-Based Agents
- G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems
- Graphs Meet AI Agents: Taxonomy, Progress, and Future Opportunities
- DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents
- MCP-Zero: Active Tool Discovery for Autonomous LLM Agents
- Simple Prompt Injection Attacks Can Leak Personal Data Observed by LLM Agents During Task Execution
- The Automated but Risky Game: Modeling and Benchmarking Agent-to-Agent Negotiations and Transactions in Consumer Markets
- Debate-to-Detect: Reformulating Misinformation Detection as a Real-World Debate with Large Language Models
- LLM-Powered AI Agent Systems and Their Applications in Industry
- AutoData: A Multi-Agent System for Open Web Data Collection
- Pierce the Mists, Greet the Sky: Decipher Knowledge Overshadowing via Knowledge Circuit Analysis
- LAMP: Extracting Locally Linear Decision Surfaces from LLM World Models
- Automated Meta Prompt Engineering for Alignment with the Theory of Mind
- Social Theory Should Be a Structural Prior for Agentic AI: A Formal Framework for Multi-Agent Social Systems
- Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use
- Characterizing AI Agents for Alignment and Governance
- GraphBit: A Graph-based Agentic Framework for Non-Linear Agent Orchestration
- A Vision for Auto Research with LLM Agents
- BackdoorAgent: A Unified Framework for Backdoor Attacks on LLM-based Agents
- Evolution of AI in Education: Agentic Workflows
- A Framework for Testing and Adapting REST APIs as LLM Tools
- From Economic Agents to Agentic Economies: A Systems Blueprint for Economic World Models
- Dual Engines of Thoughts: A Depth-Breadth Integration Framework for Open-Ended Analysis
- SynWorld: Virtual Scenario Synthesis for Agentic Action Knowledge Refinement
- Survey and Experiments on Mental Disorder Detection via Social Media: From Large Language Models and RAG to Agents
Discussions
- Survey Study on AI Agent Architectures (2024) [hn, 77 points, 16 comments]
- Survey Study on AI Agent Architectures (2024) (arxiv.org) Main Link | Discussion [bsky, 0 points, 0 comments]
- Survey on AI agent advancements, focus on reasoning, planning, and tool execution. Capabilities, observations, and design for future AI agents, discussing single and multi-agent architectures, leaders [bsky, 0 points, 0 comments]
Related