Jul 2026· 2026 6th International Conference on Electrical, Computer and Energy Technologies (ICECET)· pp. 1-6· 0 citations· 40 references
Abstract
Large Language Models (LLMs) offer a natural interface for smart-IoT control, yet reliable deployment requires more than producing valid API calls. Multi-device commands often contain preconditions, ordering constraints, and conflicts that must be satisfied before actions are safely executed. This paper presents DARIO, a dependency-aware orchestration framework for translating natural-language commands into verified IoT action plans. DARIO combines supervised instruction tuning, KL-regularized PPO, an explicit dependency graph $\mathcal{G}=(V, E)$, and a lightweight verifier that checks schema, execution, dependency, and safety constraints before commit. Verifier outcomes are folded into a decomposed reward, enabling the policy to learn from dependency violations rather than treating plans as flat token sequences. On a 2k-prompt held-out HA-Instruct split across five seeds, DARIO achieves $0.85 \pm 0.01$ JSON exact match, perfect schema validity, $0.91 \pm 0.01$ overall task success, and $0.81 \pm 0.02$ success on the dependency-heavy L3 split, outperforming SIT and SIT+PPO baselines by large margins. It also raises dependency satisfaction to 0.94, reduces unsafe execution to 0.02, and runs as a 4-bit 8B LoRA deployment at about 609 ms including verification.
Dynamic agent harnesses let language models change the software that shapes their own execution. This flexibility brings a new reasoning burden: a local plugin change can propagate through dependencies and cleanup. We introduce CordisBench, a 1,200-question benchmark of this lifecycle reasoning. It combines a controlled formal setting with programs executed against Cordis, a runtime that manages component dependencies and cleanup, and asks models to identify affected components, predict state after a specified teardown order, determine which conditions hold under all or some orders, and choose reconfigurations that succeed when executed. Across these tasks, we evaluate three efficiency-oriented models at low reasoning effort with 2, 4, 8, 16, 24, or 32 relevant interactions, using deterministic task-specific scoring. Models usually handle small systems well but grow less reliable as more interactions become relevant, especially when predicting final state and when reasoning across teardown orders. Additional inference effort recovers marked gains for some models. The cost is nontrivial: on our 16-interaction subset, GPT-5.6 Luna uses nearly 3,000 reasoning tokens per question at medium effort. For these controlled instances, that cost is avoidable: an independent finite reference semantics agrees with Cordis execution on every observation and action outcome used for scoring across all 528 executable questions.
It is shown that LLM-driven agents can violate this condition and introduced a canonical deployment wrapper that guarantees it for arbitrary base agents while preserving already-equivariant behaviour, and it is proved that computing canonical representations required by this construction is graph-isomorphism-hard.
The rapid expansion of capabilities in Large Language Model (LLM) agents has exposed a critical architectural bottleneck: when agents are given access to a flat, monolithic registry of tools, the model must evaluate hundreds or thousands of options simultaneously. This leads to decision-space explosion, context window saturation, and degraded routing accuracy. To address these limitations, this paper presents a hierarchical, skill-based architecture for agentic orchestration. Capabilities are organized as a rooted tree where internal nodes make routing decisions and leaf nodes execute deterministic tasks. The runtime enforces a single-step execution loop governed by a Last-In-First-Out (LIFO) stack, giving the agent a form of memory akin to a Pushdown Automaton, therefore enabling it to track nested execution contexts and resume deterministically from any depth. Capability discovery follows a manifest-driven, lazy-loading protocol: only the immediate children of the active node are loaded, so memory and prompt costs scale with the explored path rather than the global registry. By replacing global memory with localized stack frames, the architecture prevents outputs from one execution branch from leaking into another, establishing the isolation guarantees required for deployment in regulated enterprise environments. We also discuss UPI Help, an AI-powered digital payments support product, as a motivating production deployment context. We provide a mathematical formalization of the orchestration state, detailed algorithmic analysis of the execution loop, and controlled benchmarks comparing flat and hierarchical routing under increasing tool catalogs, multi-step workflow pressure, and visible schema-token exposure per LLM call.
Prashant Devadiga, Abhishek, Adithya Mishra et al.· 0 citations
This research paper proposes LLM-RAG Automation: Intelligent Function Execution and Secure System Orchestration, an intelligent automation solution that bridges the gap between natural language processing and autonomous executable workflows.
Avanish Cowkur, Rohan Nambiar, Rishi Nambiar et al.· International Journal of Inn...· 0 citations
This work presents FAVA (Formal Authorization for Verified Agents), a permission-carrying authorization framework for agent execution that utilizes an LLM-guided Permission Intermediate Representation to translate ambiguous natural-language tasks into structured constraints.
Yifan Zhang, Xinkui Zhao, Sai-Da Liu et al.· 3 citations