deterministic orchestration

This work addresses the fragmentation in existing frameworks that treat deterministic and probabilistic computations in isolation, lacking a unified declarative language to orchestrate large language models (LLMs) and symbolic tools. We propose Structured Prompt Language (SPL), the first framework to deeply integrate probabilistic operations (GENERATE/EVALUATE) and deterministic reasoning (SOLVE/ASSERT) within a single declarative paradigm. SPL supports shared variable binding, runtime dynamic routing, and seamless interoperability with LLMs (e.g., Ollama, Anthropic), symbolic engines (e.g., SymPy, SageMath, Lean), and the distributed execution grid Momagrid. Across 1,200 experiments, SPL achieves machine-verified correctness rates of 82–93% (e.g., 93% for gemma4:e2b), substantially outperforming pure LLM baselines; most failures stem from solver kernels rejecting invalid expressions.

This study addresses the challenges of COBOL system modernization—namely, the scarcity of domain experts, the scale of legacy codebases, and stringent correctness requirements—and investigates the efficacy of large language model (LLM)-based orchestration strategies in automated COBOL-to-Python translation. For the first time, orchestration strategy is isolated as the sole variable within a unified experimental framework, enabling a direct comparison between deterministic and LLM-driven approaches. The findings reveal that deterministic orchestration achieves functional correctness on par with LLM-based control while significantly enhancing robustness, reducing inter-run performance variability, and cutting token consumption by up to 3.5×. These results demonstrate that deterministic orchestration offers superior stability and cost efficiency without compromising translation accuracy.

This work addresses the challenges faced by large language model (LLM) agents operating over flat tool registries—namely, combinatorial explosion in decision space, context saturation, and degraded routing accuracy. To overcome these limitations, the authors propose a skill-tree-based hierarchical architecture that separates routing logic at internal nodes from execution at leaf nodes. Inspired by pushdown automata, the framework incorporates a LIFO stack-frame memory model and a lazy capability discovery mechanism, enabling isolated execution paths and scalable context management. The approach supports manifest-driven single-step execution loops and formal state modeling, significantly improving routing accuracy while reducing memory footprint and prompt costs under conditions of tool proliferation, multi-step workflows, and prompt exposure. This design meets enterprise-grade requirements for isolation and scalability.

This study addresses the challenge of automating workflows in complex industries—such as logistics, healthcare, and construction—where processes are fragmented across heterogeneous tools and involve multi-party collaboration. The work proposes orchestration as a core abstraction to enable effective automation by dynamically coordinating multi-step tasks, enforcing domain-specific constraints, managing human approvals, and integrating legacy systems. It introduces the novel concept of “orchestration bottlenecks” and develops a theoretical framework that unifies multi-agent systems, workflow modeling, constraint reasoning, and human–AI collaboration, while exposing critical gaps in current multi-agent approaches at the orchestration level. Based on distinct sources of operational friction across domains, the paper advocates for targeted architectural safeguards—such as constraint enforcement or explainability—and phased implementation strategies to provide actionable pathways for automation in complex operational environments.

This work addresses the reliability challenges of production-grade large language model (LLM) agents, which stem from the lack of a clear architectural abstraction delineating stochastic outputs from deterministic system behavior. To bridge this gap, the paper introduces the Stochastic-Deterministic Boundary (SDB) as a core architectural primitive, formalized as a four-tuple contract. Centered on three key concerns—coordination, state, and control—it defines six composable runtime modes. The contributions include a five-step methodology for mode selection, a replay-based divergence diagnosis mechanism, and the insight that architectural momentum becomes critical for long-term reliability once model variance diminishes. By integrating distributed systems patterns such as Saga and event-driven orchestration, the authors construct a verifiable, rollback-capable, and monitorable LLM agent runtime. Empirical validation across five real-world workloads demonstrates its efficacy, and a reference implementation for a 90-day contract renewal agent is open-sourced, significantly enhancing sustained operational reliability.

This work addresses the limitation of existing benchmarks, which focus solely on accuracy in multi-agent orchestration tasks while neglecting fine-grained diagnosis of failure origins and recovery capabilities. The authors propose a reproducible fault-injection framework to systematically evaluate failure modes, task decomposition quality, and recovery mechanisms within templated enterprise workflows. They introduce two novel metrics: “cascade radius” and failure-mode-specific recovery rates, and employ controlled probes to analyze recovery behavior across different fault types. Experimental results demonstrate that intent-based reasoning routing achieves 100% recovery under adversarial conditions, significantly outperforming keyword-based routing; tool-related failures are fully recoverable, whereas semantic failures prove largely irrecoverable; and cascade radius increases with workflow depth.

Reinforcement learning (RL) has long remained absent from real-world service orchestration deployments, commonly attributed to telemetry latency, load fluctuations, and tenant-induced unpredictability—yet systematic empirical validation is lacking. This study rigorously evaluates performance degradation of three representative RL-based orchestration systems under production-grade perturbations, employing preregistered experiments, paired statistical inference, and family-wise error correction. Results reveal that most performance gains claimed in the literature either fail to replicate or diminish substantially; only one system consistently outperforms Kubernetes Horizontal Pod Autoscaler (HPA) by approximately 40× under observational delay. The findings indicate that existing advantages often stem from inadequate baselines, limited artifacts, or evaluation biases, underscoring an urgent need for deployment-oriented evaluation standards and institutional incentives to bridge the gap between research and practice.

Existing evaluation methods struggle to disentangle the quality of task orchestration in multi-agent systems from confounding factors such as agent capabilities and environmental noise, while real-world execution incurs prohibitive costs. To address this, this work proposes OrchBench—a deterministic simulation-based benchmarking platform that models task dependencies via directed acyclic graphs and enables isolated, efficient assessment of orchestration plans. OrchBench achieves the first interpretable evaluation of orchestration quality with dramatically reduced overhead: requiring only 1.3% of the tokens and 10.3% of the time compared to real execution, while maintaining high fidelity (Pearson r = 0.816). Furthermore, it reveals that information retention rate is more critical to performance than simply increasing the number of agents.

In institutional settings, agent workflows may lose credibility even when producing correct outputs if they rely on faulty authorities, lack evidence of completion, or fail to respond to changes. This work proposes a “governed execution” framework that innovatively introduces a Matrix causal state layer, integrating for the first time authority dependency tracking, factual provenance, and selective invalidation mechanisms within agent workflows. By leveraging deterministic causal modeling, dependency tracing, completion verification, and precise invalidation of affected tasks, the approach ensures auditability and independent verifiability. Experiments demonstrate that governed workflows maintain result consistency while persistently preserving governance evidence, rejecting unjustified closed-loop reasoning, and constraining recovery scope; however, strict integrity constraints can lead to excessive blocking in role-separation transfer tasks.