Agentic AI: A Review of Autonomous AI Agents

References

[1] S. Yao et al., "ReAct: synergizing reasoning and acting in language models," in Proc. Int. Conf. Learning Representations (ICLR), 2023.

[2] S. Yao et al., "Tree of thoughts: deliberate problem solving with large language models," in Proc. Neural Information Processing Systems (NeurIPS), 2023.

[3] N. Shinn et al., "Reflexion: language agents with verbal reinforcement learning," in Proc. Neural Information Processing Systems (NeurIPS), 2023.

[4] J. Wei et al., "Chain-of-thought prompting elicits reasoning in large language models," in Proc. Neural Information Processing Systems (NeurIPS), 2022.

[5] T. Schick et al., "Toolformer: language models can teach themselves to use tools," in Proc. Neural Information Processing Systems (NeurIPS), 2023.

[6] Q. Wu et al., "AutoGen: enabling next-gen LLM applications via multi-agent conversation," arXiv preprint arXiv:2308.08155, 2023.

[7] J. Fourney et al., "CrewAI: a framework for orchestrating role-playing autonomous AI agents," CrewAI Documentation, 2024.

[8] H. Chase, "LangGraph: building stateful, multi-actor applications with LLMs," LangChain Documentation, 2024.

[9] T. R. Sumers et al., "Cognitive architectures for language agents," Transactions on Machine Learning Research, 2024.

[10] R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction, 2nd ed. Cambridge, MA, USA: MIT Press, 2018.

[11] R. E. Fikes and N. J. Nilsson, "STRIPS: a new approach to the application of theorem proving to problem solving," Artificial Intelligence, vol. 2, no. 3-4, pp. 189-208, 1971.

[12] P. Lewis et al., "Retrieval-augmented generation for knowledge-intensive NLP tasks," in Proc. Neural Information Processing Systems (NeurIPS), 2020.

[13] K. Cobbe et al., "Training verifiers to solve math word problems," arXiv preprint arXiv:2110.14168, 2021.

[14] L. Wang et al., "A survey on large language model based autonomous agents," Frontiers of Computer Science, vol. 18, no. 6, 2024.

[15] Z. Xi et al., "The rise and potential of large language model based agents: a survey," Science China Information Sciences, vol. 68, 2025.

[16] N. Carlini et al., "Are aligned neural networks adversarially aligned?," in Proc. Neural Information Processing Systems (NeurIPS), 2023.

[17] K. Greshake et al., "Not what you've signed up for: compromising real-world LLM-integrated applications with indirect prompt injection," in Proc. ACM Workshop on Artificial Intelligence and Security, 2023.

[18] OpenAI, "Function calling and structured outputs," OpenAI API Documentation, 2024.

[19] Anthropic, "Tool use with Claude," Anthropic API Documentation, 2024.

[20] Y. Shen et al., "HuggingGPT: solving AI tasks with ChatGPT and its friends in Hugging Face," in Proc. Neural Information Processing Systems (NeurIPS), 2023.

[21] S. Zhou et al., "WebArena: a realistic web environment for building autonomous agents," in Proc. Int. Conf. Learning Representations (ICLR), 2024.

[22] C. E. Jimenez et al., "SWE-bench: can language models resolve real-world GitHub issues?," in Proc. Int. Conf. Learning Representations (ICLR), 2024.

[23] G. Mialon et al., "GAIA: a benchmark for general AI assistants," in Proc. Int. Conf. Learning Representations (ICLR), 2024.

@article{1719864,
      author = {Rinku, Aryan, Juned, Paras},
      title = {Agentic AI: A Review of Autonomous AI Agents},
      journal = {Iconic Research And Engineering Journals},
      year = {2025},
      volume = {8},
      number = {11},
      pages = {2683-2689},
      issn = {2456-8880},
      url = {https://www.irejournals.com/formatedpaper/1719864.pdf},
      abstract = {Autonomous artificial intelligence (AI) agents, systems capable of perceiving an environment, reasoning about goals, planning multi-step actions, invoking external tools, and adapting based on feedback, have emerged as a dominant paradigm for extending the capabilities of large language models (LLMs) beyond single-turn text generation. Termed agentic AI, this paradigm encompasses single-agent architectures with planning and memory modules as well as multi-agent systems in which specialized agents collaborate, negotiate, or compete to accomplish complex objectives. This review surveys the foundations of agentic AI, including planning strategies such as ReAct, Tree-of-Thoughts, and reflection-based self-correction; memory architectures spanning short-term context windows, vector-based long-term memory, and episodic memory stores; and tool-usage frameworks that allow agents to call external application programming interfaces (APIs), execute code, and query databases. We examine prominent multi-agent frameworks including LangGraph, CrewAI, and AutoGen, comparing their orchestration models, communication protocols, and suitability for different task structures. We then survey applications of agentic AI in healthcare, finance, and education, highlighting both demonstrated benefits and domain-specific risks. Finally, we discuss open challenges including reliability and hallucination propagation across agent chains, coordination overhead in multi-agent systems, security vulnerabilities such as prompt injection, and the evaluation gap for long-horizon autonomous behavior, before outlining directions for future research.},
      keywords = {Agentic AI, Autonomous Agents, Multi-Agent Systems, LLM Planning, Tool Use, Memory Architectures, Langgraph, Crewai, Autogen, AI Safety.},
      month = {May},
  }