Architecting Agentic AI Systems: Product and System Design Patterns for Trustworthy Autonomous Decision-Making
DOI:
https://doi.org/10.22399/ijcesen.5064Keywords:
agentic AI, autonomous decision-making, trustworthy AI, LLM agents, multi-agent systems, explainable AIAbstract
The development of agentic artificial intelligence (AI) systems with the capability to perceive environments, plan, and execute multi-step tasks is a paradigmatic change in the deployment of computational intelligence. The paper has offered a synthesis of product and system design patterns that apply to trustworthy agentic AI based on the progress of large language model (LLM)-based agents, deep reinforcement learning (RL), explainable AI (XAI), fairness-aware machine learning, and governance-focused frameworks. The use of agentic AI both by enterprises and for personal purposes has expanded to around 5 per cent. in the year 2019, and is projected to grow to around 73 per cent. by the mid-2025 years, accompanied by a corresponding growth in the number of safety incidents. The scores in trustworthiness dimension are 1532 percent higher in fairness, robustness, and privacy indicators in hybrid agentic architectures in comparison with the purely LLM-based settings. Seven trustworthy AI pillars are safety, robustness, explainability, fairness, privacy, accountability, and transparency, which are aligned to the system layers with a particular design pattern. A framework namedTRiSM (Trust, Risk, and Security Management) has been found as a systematic route to the operationalization of these principles in production deployments, with 94 percent of agent impersonation incidents being reduced.
References
[1] Y. Alufaisan, L. R. Marusich, J. Z. Bakdash, Y. Zhou, and M. Kantarcioglu, “Does explainable artificial intelligence improve human decision-making?” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, no. 9, pp. 6618–6625, 2020, doi: 10.1609/aaai.v34i09.6600.
[2] S. Arya, B. K., L. Addepalli, V. S. S. D., J. Lloret, and B. Maloth, “Governance-centric framework for trustworthy and ethical autonomous decision-making in organizational systems,” in Proc. 1st Int. Conf. Research and Development in Information, Communication, and Computing Technologies (ICRDICCT’25), vol. 3, pp. 184–192, 2025, doi: 10.5220/0013893900004919.
[3] S. Baker and W. Xiang, “Explainable AI is responsible AI: How explainability creates trustworthy and socially responsible artificial intelligence,” arXiv preprint, 2023, doi: 10.48550/arXiv.2312.01555.
[4] L. Floridi et al., “AI4People—An ethical framework for a good AI society: Opportunities, risks, principles, and recommendations,” Minds and Machines, vol. 28, no. 4, pp. 689–707, 2018, doi: 10.1007/s11023-018-9482-5.
[5] V. François-Lavet, P. Henderson, R. Islam, M. G. Bellemare, and J. Pineau, “An introduction to deep reinforcement learning,” Foundations and Trends® in Machine Learning, vol. 11, no. 3–4, pp. 219–354, 2018, doi: 10.1561/2200000071.
[6] A. K. M. B. Haque, M. Rahman, and M. S. Rahman, “Explainable artificial intelligence (XAI) from a user perspective: A systematic literature review and research agenda,” Technological Forecasting and Social Change, vol. 188, Art. no. 122140, 2023, doi: 10.1016/j.techfore.2022.122140.
[7] D. Kowald, N. Rekabsaz, E. Lex, and M. Schedl, “Establishing and evaluating trustworthy artificial intelligence: A systematic overview of requirements and evaluation methods,” AI and Ethics, vol. 4, no. 3, pp. 743–759, 2024, doi: 10.1007/s43681-023-00339-4.
[8] H. Liu et al., “Trustworthy AI: A computational perspective,” ACM Transactions on Intelligent Systems and Technology, vol. 13, no. 1, Art. no. 1, pp. 1–34, 2021, doi: 10.1145/3442188.
[9] N. Mehrabi, F. Morstatter, N. Saxena, K. Lerman, and A. Galstyan, “A survey on bias and fairness in machine learning,” ACM Computing Surveys, vol. 54, no. 6, Art. no. 115, 2021, doi: 10.1145/3457607.
[10] S. Naihin et al., “Testing language model agents safely in the wild,” arXiv preprint, 2023, doi: 10.48550/arXiv.2311.10538.
[11] J. S. Park, J. C. O’Brien, C. J. Cai, M. R. Morris, P. Liang, and M. S. Bernstein, “Generative agents: Interactive simulacra of human behavior,” arXiv preprint, 2023, doi: 10.48550/arXiv.2304.03442.
[12] Z. Pei, Y. Zhang, and H. Chen, “Reinforcement learning for decision-making under deep uncertainty,” Journal of Environmental Management, vol. 351, Art. no. 120968, 2024, doi: 10.1016/j.jenvman.2024.120968.
[13] S. Raza, R. Sapkota, M. Karkee, and C. Emmanouilidis, “TRiSM for agentic AI: A review of trust, risk, and security management in LLM-based agentic multi-agent systems,” arXiv preprint, 2025, doi: 10.48550/arXiv.2506.04133.
[14] R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction, 2nd ed. Cambridge, MA, USA: MIT Press, 2018, doi: 10.7551/mitpress/10955.001.0001.
[15] C. Wu, Y. Zhao, and Z. Liu, “Survey of trustworthy artificial intelligence: Foundations and future research directions,” arXiv preprint, 2023, doi: 10.48550/arXiv.2306.00380.
[16] Z. Xi et al., “The rise and potential of large language model based agents: A survey,” arXiv preprint, 2023, doi: 10.48550/arXiv.2309.07864.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2025 International Journal of Computational and Experimental Science and Engineering
This work is licensed under a Creative Commons Attribution 4.0 International License.
