Citation :

Ashok Kumar, Dhruv Kumar Seth, Karan Kumar Ratra,Vikas Kumar Mittal, "AI-Based Intent Normalization and Service Routing in Distributed Agentic Microservices," International Journal of Computer Trends and Technology (IJCTT), vol. 74, no. 9, pp. 25-33, 2026. Crossref, https://doi.org/10.14445/22312803/IJCTT-V74I9P103

Abstract

LLMs are increasingly being embedded into a variety of distributed agentic systems, in the form of agents with APIs, microservices, workflows, and external tools for task completion. Systems need to be able to handle requests that are worded differently but mean the same thing, to make sure routing is consistent, caching can be reused, latency is reduced, reliability is ensured, and governance is supported. Conventional intent representations, beyond simply conveying what users want with their intent and respective slots, do not encode information about user authorization, policies with implications for execution, side effects of actions, locality of execution, and operational implications of intents. Prior work has provided frameworks for normalization, capability discovery and routing in distributed agentic systems, but all this work is done independently and thus does not provide a unified pipeline for handling requests in distributed systems. The main contribution of this paper is to propose a unified framework describing the normalization, discovery, routing, execution control and runtime feedback in distributed agentic systems. The core of this work is the introduction of a canonical representation of intent, called an intent record, which can convey all kinds of semantic information, as well as context, policy, and operational information about the system as a whole, including its uncertainty, to be used by the system’s routing layer. We also introduce a new constraint-first multi-criteria routing process. A reference architecture is also given for such sorts of systems. The architecture’s main functional components are semantic control and its decoupled, cross-governance-deterministic execution. An evaluation framework for the proposed approach in the future is also presented. This framework is not validated through implementation or benchmarking experiments. Such an empirical evaluation is left for future work.

Keywords

Intent Normalization, Service Routing, Agentic Microservices, Canonical Intent Representation, Distributed Systems, Multi-Agent Orchestration.

References

[1] Fu Bang, “GPTCache: An Open-Source Semantic Cache for LLM Applications Enabling Faster Answers and Cost Savings,” Proceedings of the 3^rd Workshop for Natural Language Processing Open Source Software (NLP-OSS 2023), pp. 212-218, 2023.
[CrossRef] [Google Scholar] [Publisher Link]

[2] Qian Chen, Zhu Zhuo, and Wen Wang, “BERT for Joint Intent Classification and Slot Filling,” arXiv, pp. 1-6, 2019.
[CrossRef] [Google Scholar] [Publisher Link]

[3] Jonathan Berant, and Percy Liang, “Semantic Parsing via Paraphrasing,” Proceedings of the 52^nd Annual Meeting of the Association for Computational Linguistics (ACL), pp. 1415-1425, 2014.
[CrossRef] [Google Scholar] [Publisher Link]

[4] Ehud Karpas et al., “MRKL Systems: A Modular, Neuro-Symbolic Architecture that Combines Large Language Models, External Knowledge Sources and Discrete Reasoning,” arXiv, pp. 1-19, 2022.
[CrossRef] [Google Scholar] [Publisher Link]

[5] Shunyu Yao et al., “ReAct: Synergizing Reasoning and Acting in Language Models,” International Conference on Learning Representations (ICLR), pp. 1-33, 2023.
[Google Scholar] [Publisher Link]

[6] Timo Schick et al., “Toolformer: Language Models Can Teach Themselves to use Tools,” Advances in Neural Information Processing Systems, Neural Information Processing Systems Foundation, Inc. (NeurIPS), vol. 36, pp. 1-13, 2023.
[CrossRef] [Google Scholar] [Publisher Link]

[7] Yongliang Shen et al., “HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Face,” Advances in Neural Information Processing Systems, vol. 36, pp. 1-27, 2023.
[Google Scholar] [Publisher Link]

[8] Qingyun Wu et al., “AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation,” arXiv, pp. 1-43, 2023.
[CrossRef] [Google Scholar] [Publisher Link]

[9] Minghao Li et al., “API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs,” Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, Association for Computational Linguistics, pp. 3102-3116, 2023.
[CrossRef] [Google Scholar] [Publisher Link]

[10] Yujia Qin et al., “ToolLLM: Facilitating Large Language Models to Master 16000+ Real-World APIs,” arXiv, pp. 1-24, 2023.
[CrossRef] [Google Scholar] [Publisher Link]

[11] Shishir G. Patil et al., “Gorilla: Large Language Model Connected with Massive APIs,” arXiv, pp. 1-18, 2024.
[CrossRef] [Google Scholar] [Publisher Link]

[12] Fanjia Yan et al., Berkeley Function-Calling Leaderboard, Berkeley Sky Computing Lab, 2024. [Online]. Available: https://gorilla.cs.berkeley.edu/blogs/8_berkeley_function_calling_leaderboard.html

[13] Zhicheng Guo et al., “StableToolBench: Towards Stable Large-Scale Benchmarking on Tool Learning of LLMs,” arXiv, pp. 1-14, 2024.
[CrossRef] [Google Scholar] [Publisher Link]

[14] Guozhao Mo et al., “LiveMCPBench: Can Agents Navigate an Ocean of MCP Tools?,” arXiv, pp. 1-18, 2025.
[CrossRef] [Google Scholar] [Publisher Link]

[15] Zhenting Wang et al., “MCP-Bench: Benchmarking Tool-using LLM Agents with Complex Real-World Tasks via MCP Servers,” arXiv, pp. 1-52, 2025.
[CrossRef] [Google Scholar] [Publisher Link]

[16] Xuanqi Gao et al., “MCP-RADAR: A Multi-Dimensional Benchmark for Evaluating Tool use Capabilities in Large Language Models,” arXiv, pp. 1-20, 2025.
[CrossRef] [Google Scholar] [Publisher Link]

[17] Chaithanya Bandi et al., “MCP-Atlas: A Large-Scale Benchmark for Tool-Use Competency with Real MCP Servers,” arXiv, pp. 1-25, 2026.
[CrossRef] [Google Scholar] [Publisher Link]

[18] Anthropic, Introducing the Model Context Protocol, 2024. [Online]. Available: https://www.anthropic.com/news/model-context-protocol

[19] Model Context Protocol, Specification, 2025. [Online]. Available: https://modelcontextprotocol.io/specification/2025-03-26

[20] Model Context Protocol, Authorization, 2026. [Online]. Available: https://modelcontextprotocol.io/specification/2025-03-26/basic/authorization

[21] Google, Announcing the Agent2Agent Protocol (A2A), Google Developers Blog, 2025. [Online]. Available: https://developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/

[22]  Kubernetes SIG Network, Gateway API: Introduction. [Online]. Available: https://gateway-api.sigs.k8s.io/docs/introduction/

[23]  OpenTelemetry, Documentation. [Online]. Available: https://opentelemetry.io/docs/

[24] World Wide Web Consortium, Trace Context, W3C Recommendation. [Online]. Available: https://www.w3.org/TR/trace-context