Open research questions in Multi-Agent Systems and Negotiation

What the literature leaves open

• against predefined business policies, ensuring that generated actions satisfied margin protection and pricing constraints before execution,. evidence, provided the Explainability Framework Furthermore, supporting information, contributing confidence scores, and governance validation results for every transparency of improved recommendation. This autonomous decision making and increased the interpretability of pricing recommendations from both merchant and customer perspectives,. agent the As the current implementation focuses on architectural and functional validation, the evaluation emphasizes successful workflow execution, system integration, and operational behavior rather than large-scale quantitative performance metrics. Nevertheless, the prototype demonstrates that the proposed architecture is technically feasible and suitable for future enterprise-scale deployment and empirical evaluation. 7.3 Discussion The experimental evaluation demonstrates that integrating multiple enterprise intelligence modules within a governanceaware multi-agent architecture is both practical and operationally feasible. Unlike conventional pricing systems isolated business objectives, BazaarAI that optimize intelligence, pricing optimization, coordinates customer VII. EXPERIMENTAL EVALUATION 7.1 Prototype Implementation The proposed BazaarAI framework was implemented as a functional enterprise SaaS prototype to validate the feasibility of the proposed governance-aware multi-agent architecture. The prototype supports end-to-end pricing and negotiation workflows interface and demonstrates the integration of Customer Intelligence, Pricing Intelligence, Revenue Intelligence, Opportunity Detection, Governance, Explainable AI, and Multi-Agent Collaboration within a unified decision environment,,. interactive web through an pricing recommendations, The implementation includes a representative product catalog and multiple enterprise pricing scenarios covering customer negotiation, inventory-aware decision making, and governance validation. Twelve specialized AI agents collaborate to analyze customer behavior, business constraints, and pricing conditions before generating explainable recommendations,. Every recommendation is validated against predefined governance policies, including margin protection, discount thresholds, and pricing constraints, before execution,. The prototype also provides enterprise dashboards, an Agent Operations Center, a pricing simulator, and an enterprise integration layer, demonstrating how BazaarAI can be integrated with existing commerce platforms through APIs rather than operating as a standalone application. Figures 1–9 illustrate the major architectural components and enterprise interfaces implemented within the prototype. © 2026 The Author(s). Published by IJCOPE Journal. Website: https://ijcope.org/ 10 International Journal of Creative and Open Research in Engineering and Management ISSN: 3108-1754 (Online) Volume 02 Issue 07 Jul-2026 | Impact Factor: 3.5 revenue analysis, governance enforcement, and explainability within a unified decision framework,,. The modular architecture enables flexible integration with existing enterprise systems such as e-commerce platforms, CRM solutions, ERP systems, and inventory management applications through standardized APIs. This integrationoriented design makes BazaarAI suitable for enterprise environments where pricing intelligence must operate alongside existing business infrastructure.

  BazaarAI: A Governance-Aware Multi-Agent Framework for Intelligent Pricing and Nego · 2026 · DOI

• that violate predefined rules may be modified, rejected, or escalated for manual review. This governance-first approach improves trust, accountability, and enterprise readiness while reducing operational risk. 4.6 Learning Agent The Learning Agent enables continuous adaptation and longterm system improvement. Following action execution, it collects performance feedback and compares actual outcomes with predicted results to refine future recommendations. Key monitoring areas include: • Revenue Outcomes • Conversion Performance • Margin Realization • Negotiation Success Rates © 2026 The Author(s). Published by IJCOPE Journal. Website: https://ijcope.org/ 7 International Journal of Creative and Open Research in Engineering and Management ISSN: 3108-1754 (Online) Volume 02 Issue 07 Jul-2026 | Impact Factor: 3.5 • Customer Retention Indicators • Promotional Effectiveness • Decision Accuracy By analyzing historical performance and emerging business patterns, the Learning Agent continuously improves system effectiveness while maintaining governance compliance and operational stability. Its inclusion transforms BazaarAI from a static decision-support tool into a self-improving revenue decision platform. 4.7 Consensus Mechanism The Consensus Mechanism serves as the central coordination component of the multi-agent framework and is a key contribution of BazaarAI. Enterprise objectives often conflict with one another. For example, discounting may increase conversions but reduce profitability, while strict margin controls may limit customer acquisition. Instead of optimizing a single objective, BazaarAI aggregates recommendations from multiple specialized agents and evaluates them through a collaborative decision process. Each agent contributes: • Recommended Action • Confidence Score • Expected Business Impact • Supporting Evidence • Risk Assessment The Consensus Mechanism combines these inputs to identify actions that best balance customer satisfaction, profitability, revenue growth, inventory utilization, and governance compliance. A conceptual utility function is defined as: U = w₁R + w₂M + w₃C + w₄I + w₅L where: R = Expected Revenue Impact M = Margin Preservation C = Conversion Probability I = Inventory Health L = Customer Lifetime Value The weighting factors are determined according to organizational priorities and strategic objectives.Through this collaborative optimization process V. Governance-Aware Decision Engine The Governance-Aware Decision Engine acts as the control layer of BazaarAI, ensuring that all revenue decisions comply with organizational policies, profitability requirements, and operational constraints. Rather than treating governance as a post-processing step, BazaarAI embeds governance directly into the decision-making process. This governance-centric approach enables autonomous yet accountable optimization while maintaining transparency, consistency, and enterprise trust.

  BazaarAI: A Governance-Aware Multi-Agent Framework for Intelligent Pricing and Nego · 2026 · DOI

• conversion, inventory utilization, and long-term growth. To address this challenge, BazaarAI adopts a collaborative multiagent architecture in which specialized agents contribute domain-specific intelligence while collectively supporting decision-making. Each agent analyzes a specific business dimension are consolidated through a consensus mechanism before final action scalability, modularity, explainability, and adaptability by decomposing complex revenue decisions into coordinated intelligence tasks. 5.1 Customer Agent The Customer Agent represents customercentric objectives within the decision-making process. Its primary goal is to maximize customer engagement, conversion probability, and long-term value The agent analyzes behavioral signals such as browsing activity, purchase intent, customer lifetime value, churn risk, engagement metrics, and price sensitivity.

  BazaarAI: A Governance-Aware Multi-Agent Framework for Intelligent Pricing and Nego · 2026 · DOI

• systems, frameworks. Customer intelligence systems analyze behavior and preferences, pricing engines optimize prices based on market conditions, and revenue intelligence platforms support these forecasting and performance monitoring. However, technologies are typically deployed as independent systems with limited coordination. The emergence of multi-agent AI offers new possibilities for enterprise decision intelligence. Specialized agents can collaborate while focusing on distinct business objectives, enabling the integration of customer intelligence, pricing revenue optimization, intelligence, governance, and decision support within a unified framework.

  BazaarAI: A Governance-Aware Multi-Agent Framework for Intelligent Pricing and Nego · 2026 · DOI

• decision-making, explainable a The major contributions of this paper are: • A unified multi-agent architecture integrating customer, pricing, revenue, and opportunity intelligence. • A governance-aware decision framework that validates actions against business constraints. • An explainable recommendation mechanism that improves transparency and trust. • A collaborative agent-based approach for balancing revenue, margin, customer experience, and inventory objectives. • An enterprise-scale Revenue Decision Operating System learning and autonomous revenue supporting continuous optimization. 1.1 Background The rapid digitalization of commerce has transformed how organizations manage pricing, customer engagement, and revenue generation. Modern enterprises operate in competitive environments where pricing decisions must adapt to customer behavior, competitor actions, market conditions, inventory levels, and business objectives. As organizations generate increasing volumes of data, there is a growing need for intelligent systems that can convert this information into actionable business decisions. Traditional pricing approaches rely on static rules and predefined discount structures. While simple to implement, these methods often fail to respond effectively to dynamic market conditions, resulting in revenue leakage, excessive discounting, missed opportunities, and inefficient resource utilization. To overcome these limitations, increasingly adopt data-driven pricing and organizations revenue management artificial solutions powered by intelligence, machine learning, and predictive analytics. Recent advances in enterprise AI have led to the development of customer intelligence platforms, dynamic pricing engines, revenue recommendation systems, frameworks. Customer intelligence systems analyze behavior and preferences, pricing engines optimize prices based on market conditions, and revenue intelligence platforms support these forecasting and performance monitoring. However, technologies are typically deployed as independent systems with limited coordination. The emergence of multi-agent AI offers new possibilities for enterprise decision intelligence. Specialized agents can collaborate while focusing on distinct business objectives, enabling the integration of customer intelligence, pricing revenue optimization, intelligence, governance, and decision support within a unified framework.

  BazaarAI: A Governance-Aware Multi-Agent Framework for Intelligent Pricing and Nego · 2026 · DOI

• These findings suggest that agentic LLM systems can generate clinically relevant breast cancer recommendations, but remain insufficient for unsupervised clinical use.

  Agentic systems for breast cancer treatment recommendations · 2026

• Exploration is essential for reliable autonomy in multi-agent systems, yet it remains unclear whether large language model (LLM) agents can explore effectively when interacting with one another.

  Multi-Agent LLMs Fail to Explore Each Other · 2026

• Several extensions merit investigation: • Temporal, directed, and adaptive topologies. Extending the framework to time-varying adjacency and Bwrite matrices A(t) and B(t) would capture the dynamic rewiring inherent in agent swarms where new tools are registered, agents are spawned, or communication patterns shift at runtime. Furthermore, decomposing the symmetric bipartite matrix B into directional matrices ( Bread ) would enable the modeling of asymmetric access controls (e.g., read- only vs. write-only tool permissions). This introduces a weighted, directed spectral problem that may require numerical rather than analytical treatment. Continuous-time coupled formulations ODEs (Paré and Beck 2020) could complement our discrete-time MMCA in capturing dynamic permission changes over time. using • Heterogeneous transmission rates. Replacing scalar transmission rates with edge-specific parameters (e.g., βA ik ) would allow modeling of varying tool trust levels and agent privilege hierarchies. ij and βB • Game-theoretic attacker models.

  Cross-layer contagion of prompt injections in multi-agent swarms: a multiplex microscopic markov chain approach · 2026 · DOI

• Recent LLM-based mathematical reasoning agents have begun to tackle research-level problems and, in several cases, have contributed to the resolution of open problems.

  Danus: Orchestrating Mathematical Reasoning Agents with Fact-Graph Memory · 2026

• Recent frameworks clarify how attention, memory, and reasoning differences shape human-AI interaction at the individual and dyadic levels, but a formal account of how these differences scale to group-level dynamics is lacking.

  Collective Cognition in Hybrid Groups: A Network Science Synthesis · 2026

• Based on the empirical findings and deployment analysis presented in this study, several critical and actionable recommendations are proposed for engineers and system architects intending to implement production-grade centralized Agent-to-Agent (A2A) negotiation orchestration systems. First, the enforcement of a zero-trust security architecture must be established from the outset as a foundational requirement. This includes the deployment of SPIFFE/SPIRE-based workload identity frameworks utilizing short-lived X.509 Secure Verifiable Identity Documents (SVIDs), combined with mutual Transport Layer Security (mTLS) enforced through service meshes such as Istio or Linkerd, with a strict 24-hour certificate rotation policy. Furthermore, all directive bundles must be cryptographically signed using HMAC-SHA256, and any unsigned or unverifiable directives must be categorically rejected. This requirement is substantiated by ablation results demonstrating a drastic decline in security effectiveness—from 95% to 51%—when cybersecurity controls are removed under adversarial conditions. Second, robust database persistence configurations are essential to ensure system reliability and recoverability.

  Centralized Orchestration for High-Frequency Agent-to-Agent Negotiations: The Puppeteer Pattern · 2026 · DOI

• Several promising directions remain for extending the Puppeteer Pattern: • Adaptive Hybrid Architectures: Dynamic arbitration mechanisms that adaptively shift the locus of control based on real-time environmental conditions (Renting et al., 2020). • Federated and Hierarchical Deployments: Research on optimal cluster sizing, inter-tier protocols, and fault-tolerance for N > 50,000 agent populations (Hasan et al., 2024; Moore, 2025). • Learning-Based Agent Integration: How a central Puppeteer can effectively fine-tune or reward- shape decentralized learning agents and ensure interpretability of hierarchical decisions. • Adversarial Robustness: Formal performance guarantees for non-cooperative and deceptive agent behaviors in high-frequency settings. • Domain Application: Decentralized energy markets (Pinto et al., 2018), autonomous supply chain management (Bastos et al., 2023), and telecommunications bandwidth allocation.

  Centralized Orchestration for High-Frequency Agent-to-Agent Negotiations: The Puppeteer Pattern · 2026 · DOI

• Multi-agent LLM systems are increasingly deployed in settings where individual agent failures can cascade across the collective, yet the mechanisms by which such failures propagate—and the structural conditions under which they are contained or corrected—remain poorly characterized.

  Failure Propagation and Self-Correction in Multi-Agent LLM Systems: How Deliberative Consensus, Credit Assignment, and Architectural Isolation Jointly Determine Systemic Reliability · 2026 · DOI

• Yet the reliability of these judges for deep research agents remains poorly understood, posing a critical meta-evaluation problem: before deploying LLM judges to supervise research agents, we must first evaluate the judges themselves.

  Time to REFLECT: Can We Trust LLM Judges for Evidence-based Research Agents? · 2026

• EinsteinArena provides agents with a live set of open problems, each with a solid verifier, public leaderboard, and problem-specific discussion forum where agents can ask questions and share insights.

  Harnessing the Collective Intelligence of AI Agents in the Wild for New Discoveries · 2026

• Agentic AI is increasingly discussed as a prospective participant in business processes, yet it remains unclear how current implementations relate to this vision.

  Agentic AI In Business Processes: A Conceptually Grounded Analysis Of Open-Source Implementations · 2026

• Designing polynomial-time algorithms for approximate Nash equilibria (ANE) with provable worst-case guarantees is a fundamental open problem in algorithmic game theory.

  Discovering expert-level Nash equilibrium algorithms with large language models · 2026 · DOI

• We further outline open challenges for harness engineering, including evaluation beyond final task success, verification under incomplete feedback, regression-free harness improvement, consistent shared state across multiple agents, human oversight for safety-critical actions, and extensions to multimodal environments.

  Code as Agent Harness · 2026

• First formalized in the Hearsay-II speech understanding system [2] and generalized by Hayes-Roth [3] and Nii [4], blackboard ar- chitectures organize problem solving around a shared global data structure (“the blackboard”) that is updated by independent Knowl- edge Sources (KS). A control component monitors blackboard state and opportunistically activates KS modules whose preconditions are satisfied. While this approach offers flexibility, the resulting execution order can be difficult to predict and debug in operational environments. Recent LLM multi-agent frameworks revisit cooper- ative problem solving but largely through message-passing rather than shared state [6][7][8][9][10][11]; two recent works revisit the blackboard model directly [12][13] but neither targets a specific do- main with a deterministic pipeline or reports a running operational system with per-KS observability.

  Deterministic Blackboard Pipelines with Specialized LLM Knowledge Sources: A Generalizable Architecture for Intelligent Multi-Stage Reasoning · 2026 · DOI

• This paper has chartered the emergence of agentic AI as yet another transformative paradigm, representing a significant evolutionary leap from traditional AI agents and reactive generative AI. We demonstrated that agentic AI, characterized by their goal-oriented autonomous behavior, task decomposition and planning capabilities, and ability to orchestrate the constituent AI agents to interact with external tools, are moving AI to the role of an active problem-solver. Our primary contribution has been to bridge the gap between the high-level theory of agentic AI and its practical, high-stakes application within the field of engineering. To achieve this, we first establish a clear and necessary taxonomy, distinguishing the unique capabilities of agentic AI from its predecessors. Four state-of-the art use cases in the field of electrical engineering are presented, ranging from agentic AI based power system simulation software benchmarking and simulation studies to automated substation illumination design, automated bill-of-quantity generation, and advanced survival analysis for EV framework with the goal to identify the most suitable profit maximizing pricing strategy. However, this transformative potential is accompanied by significant risks. Our investigation into failure modes identified critical vulnerabilities, including the adversarial spread of false information and the cascading degradation of information accuracy through successive LLM rewrites. In response, we proposed tangible mitigation strategies, including the adaptation of a Zero Trust Framework (ZTF) to enforce continuous verification of agent identity and data, and a novel information clustering architecture to protect data integrity. Our technical solutions are complemented by practical recommendations for deploying trustworthy systems, emphasizing the vital role of Humanin-the-Loop (HITL), adherence to emerging AI standards, self-documented immutable audit trails for better accountability of the constituent AI agents that forms the agentic AI framework. Looking ahead, while the potential of agentic AI is clear, its robust and scalable deployment hinges on addressing several key research areas. The following are some of the high-impact research trajectories derived from the gaps we have identified from our current research: • Standardization of engineering tool integration: As highlighted in the power systems case study, the lack of official standardized Model Context Protocols (MCPs) from engineering software vendors such as PSS®E, PowerWorld, CDEGS is a major barrier for standardized deployment. Early progress is likely to be driven by community-developed MCPs, but future research must focus on developing open, stable, and secure MCPs for more robust agenttool interaction to ensure interoperability, reliability, and foster a collaborative development ecosystem.

  Agentic AI systems in electrical power systems engineering: current state-of-the-art and challenges · 2026 · DOI

• In this paper, we have presented a modification-based ap- proach to splitting assumption-based argumentation frame- works. First, we have introduced a splitting schema for SETAFs, to make splitting available on ABAFs after in- stantiation. Leveraging on the close connection between SETAFs and ABAFs, we have thus proposed a way to split when reasoning in ABA is performed indirectly via the cor- responding argument graph. Moreover, to overcome the re- quirement of an instantiation and its associated costs, we have introduced a splitting schema that works directly on ABA knowledge bases. For both approaches, we have shown that extensions of a given ABAF can be obtained incrementally from its sub-frameworks, by means of sim- ple syntactic modifications. Conversely, we can project an arbitrary extension of the whole framework to its sub- frameworks. Since this is bound to the specific structure of the underlying ABAF, we have considered a more gen- eral variant of splitting called parametrised splitting inspired by Baumann et al. (2012). Moreover, it is easy to see that each of the steps involved can be carried out efficiently and implemented on top of common ABA (or SETAF) solvers. Indeed, the splitting techniques introduced in the paper re- quire only syntactic modifications to the sub-frameworks, which do not lead to an exponential increase in size (and may, in some cases, even reduce it). In particular, through the reduct and modification, we introduce at most one ad- ditional rule and one additional assumption with respect to the original sub-framework.

  Splitting Assumption-Based Argumentation Frameworks · 2026 · DOI

• This paper introduced a framework for self-evolving software agents that integrates automated software evolution principles within a BDI architecture augmented by LLMs. By separating runtime reasoning from an explicit evolution module, the proposed approach enables agents to autonomously revise goals, reasoning structures, and executable actions, moving beyond traditional notions of behavioural adaptation. The prototype and preliminary evaluation demonstrate the feasibility of autonomous evolution in dynamic multi-agent environments, while also revealing current limitations in behavioural inheritance, stability, and scalability. These challenges point to the need for reinforcement mechanisms, memory consolidation, and more robust selection strategies. Future work will focus on strengthening inheritance and longterm consistency, extending the evaluation to more complex multiagent scenarios, and exploring collective and cooperative forms of software evolution among agents, as well as leveraging retrievalaugmented generation to enhance LLM reasoning with external knowledge. REFERENCES L. Bettini. 2015. Implementing Domain-Specific Languages with Xtext and Xtend. Packt Publishing, Birmingham, UK. B. W. Boehm. 1988. A spiral model of software development and enhancement. ACM SIGSOFT Software Engineering Notes 11, 4 (1988), 14–24. M. Böhm and A. Zimmermann. 2020. The Autonomous System Dilemma: Balancing Adaptability and Predictability. IEEE Software 37, 4 (2020), 44–49. R. et al. Bommasani. 2021. On the opportunities and risks of foundation models. arXiv preprint arXiv:2108.07258 (2021), e220119. Tom B Brown et al. 2020. Language Models are Few-Shot Learners. Advances in Neural Information Processing Systems 33 (2020), 1877–1901. J. M. Burge and D. C. Brown. 1999. Software change: Cost, causes, and complexity. Software Engineering Journal 14, 3 (1999), 180–190. Mark Chen et al. 2021. Evaluating Large Language Models Trained on Code. arXiv preprint arXiv:2107.03374 (2021). B. H. Cheng, H. Giese, P. Inverardi, and J. Magee. 2009. Software Engineering for Self-Adaptive Systems: A Research Roadmap. Software Engineering for Self- Adaptive Systems (2009), 1–26. T. H. Davenport and R. Kalakota. 2019. The potential for artificial intelligence in healthcare. Future Healthcare Journal 6, 2 (2019), 94–98. R. de Lemos, H. Giese, H. A. Müller, and M. Shaw. 2001. Self-adaptive software: Landscape and research challenges. ACM Transactions on Autonomous and Adaptive Systems 4, 2 (2001), 1–25. Juan Fernandez-Ramil, Dewayne Perry, and Nazim H. Madhavji (Eds.). 2006. Software Evolution and Feedback: Theory and Practice. Wiley, Chichester. S. Franklin and A. Graesser. 1996. Is it an agent, or just a program?: A taxonomy for autonomous agents. In Proceedings of the International Workshop on Agent Theories, Architectures, and Languages. Springer, Berlin, Heidelberg, 21–35. D.

  Self-Evolving Software Agents · 2026 · DOI

• We view AgentFuel as a first but significant step in advancing evals for timeseries analysis agents. We conclude by acknowledging several limitations and directions for improvement. First, we focus only on the analysis workflow and ignore other issues in data 9 wrangling or data cleaning. Second, we focus on one-shot agents that answer the question as posed; a natural direction is to extend AgentFuel to agents that can ask questions to refine the intent. Third, we showed the value of AgentFuel as a testing framework. As future work, we plan to provide more fine-grained debugging (e.g., possibly connecting with backend tracing) to understand failure patterns and continuous training.

  Generating Expressive and Customizable Evals for Timeseries Data Analysis Agents with AgentFuel · 2026 · DOI

• The current framework depends on the quality of embed-dings and the availability of legal datasets. External precedent retrieval may be affected by API limitations. Future work includes expanding datasets, optimizing agent communication, and integrating domain-specific fine-tuned models.

  A Multi-Agent Retrieval-Augmented Generation Framework for Context-Aware Legal Document Analysis · 2026 · DOI

• JUDGE 373 Still other kinds of coherence can and should be examined by means of research on any proposed command system; such research can range from highly informal studies of system elements to major formal simula- tions of the system as a whole.

  JUDGE: A value-judgment-based tactical command system · 1967 · DOI

Most-cited papers in Multi-Agent Systems and Negotiation

• The ODD Protocol for Describing Agent-Based and Other Simulation Models: A Second Update to Improve Clarity, Replication, and Structural Realism · Journal of Artificial Societies and Social Simulation · 2020 · 703 citations
• A survey on LLM-based multi-agent systems: workflow, infrastructure, and challenges · Vicinagearth. · 2024 · 282 citations
• Large language models empowered agent-based modeling and simulation: a survey and perspectives · Humanities and Social Sciences Communications · 2024 · 197 citations
• ExpeL: LLM Agents Are Experiential Learners · Proceedings of the AAAI Conference on Artificial Intelligence · 2024 · 105 citations
• One Theory - Many Formalizations: Testing Different Code Implementations of the Theory of Planned Behaviour in Energy Agent-Based Models · Journal of Artificial Societies and Social Simulation · 2018 · 59 citations
• Advancing the Search Frontier with AI Agents · Communications of the ACM · 2024 · 50 citations
• ‘One Size Does Not Fit All’: A Roadmap of Purpose-Driven Mixed-Method Pathways for Sensitivity Analysis of Agent-Based Models · Journal of Artificial Societies and Social Simulation · 2020 · 50 citations
• Methods That Support the Validation of Agent-Based Models: An Overview and Discussion · Journal of Artificial Societies and Social Simulation · 2024 · 50 citations
• Human-artificial interaction in the age of agentic AI: a system-theoretical approach · Frontiers in Human Dynamics · 2025 · 50 citations
• Using Machine Learning for Agent Specifications in Agent-Based Models and Simulations: A Critical Review and Guidelines · Journal of Artificial Societies and Social Simulation · 2023 · 35 citations

Most recent work

• A multi-agent system for automating scientific discovery · Nature · 2026
• The Master-Embedded Device: Transferring Tacit Knowledge Density into AI Agent Architecture · Zenodo (CERN European Organization for Nuclear Research) · 2026
• Agentic AI systems in electrical power systems engineering: current state-of-the-art and challenges · Frontiers in Artificial Intelligence · 2026
• Multi-Agent Ethnography: Post-Conventional Anthropological Practice Through Human−AI Collaboration · Anthropological Forum · 2026
• Beyond Black Boxes: Designing and Testing Agentic AI Systems for Strategy · Strategy Science · 2026
• A collaborative and sustainable engineering design framework enabled by agentic AI · International Journal of Sustainable Engineering · 2026
• Breaking Barriers: Multi-Agent Prompt Fusion for Automated LLM Jailbreaks · Cognitive Computation · 2026
• Coalition Formation Events: How Multi-Agent Systems Create Temporary Actors · Zenodo (CERN European Organization for Nuclear Research) · 2026
• From benchmarks to deployment: a comprehensive review of agentic AI evaluation · Artificial Intelligence Review · 2026
• Design and Implementation of Agentic AI Pipelines for Enterprise Decision-Making Architecture Patterns for Production Systems · International Journal of Innovative Research in Engineering & Multidisciplinary Physical Sciences · 2026

Related topics in Computer Science