Volume 7 Issue 9 2025

Large Language Models (LLMs) excel in natural language processing but face significant challenges in complex, multistep tasks requiring external tool integration and dynamic decision-making. Traditional orchestration methods, such as prompt chaining, lack flexibility, resulting in inefficiencies, poor error handling, and significant context loss. This paper proposes LangGraph, a graph-based orchestration framework built on LangChain, which models LLMs and tools as nodes in a directed acyclic graph (DAG) with conditional transition edges. LangGraph enhances scalability, fault tolerance, and context management, enabling modular, tool-aware AI workflows. Extensive experiments across diverse domains demonstrate a 92% task success rate, 30% reduction in token usage, and 85% error recovery rate compared to linear pipelines. The framework’s efficacy is evaluated in research agents, autonomous assistants, IoT workflows, and healthcare applications. LangGraph’s opensource nature and lightweight design (200 MB runtime footprint) make it a transformative solution for intelligent, adaptive systems, suitable for both cloud and edge deployments.

51% of India,s air pollution is caused by industrial pollution. 27% by vehicle, 17% by crop burning and 5% other sources. Air pollution contributes to the premature deaths of 2 million Indians every year. Air pollution in India is a serious environmental issue. 30% most polluted cities in the world. 21 were in India in 2019. As per a study based on 2016 data, at least 140 million people in India breathe air that is 10 times or more over the WHO sate limit and 13 of the World,s 20 cities with the highest annual levels of air pollution are in India. The air ( prevention and control of pollution) act was passed in 1981 to regulate air pollution but has failed to reduce pollution because of poor enforcement of the rules. India has a low per capita emissions of greenhouse gases but the country as a whole is the third largest greenhouse gas producer after China and the United states. A 2013 study on nonsmokers has found that Indians have 30% weaker lung function than Europeans. Major outdoor pollution sources include residential energy for cooking and heating, vehicles, power generation, agriculture/ waste incineration, and industry.

Purpose – This research throws light on the factors creating customer trust in AI in Indian banking sphere of public and private banks. It examines the ways in which usefulness, ease of use, security, and awareness of AI shape customer attitudes and intentions toward AI adoption, which contribute to their trust in AI. The mediating effects of attitude and behavioural intention between these perceptions and trust formation are also contemplated in investigation. Design/methodology/approach – In an attempt to confirm the model, data were gathered from 453 Indian bank customers via a structured questionnaire. PLS-SEM was administered for the evaluation of the proposed conceptual model, while MGA was conducted for comparing the responses of public sector versus private sector bank users. Findings – Perceived security turns out to be the strongest factor influencing attitude toward AI and intention to use AI, followed by ease of use and usefulness. Attitude and intention are significant mediators between technological perceptions and customer trust. Contrary to expectations, AI knowledge had no influence on trust, neither directly nor indirectly. Another set of results relates to the sectorial differences of those customers: customers in public sector emphasized security all the more. Research Limitations/Implications – The cross-sectional nature of this study constrains analysis over time, and the results may not be generalized to the population beyond digitally literate respondents in India. Future works can follow the longitudinal and qualitative approach, including more constructs like algorithmic transparency, and extending the model to other service industries. Originality/Value – The research investigates an extension of the Technology Acceptance Model through more trust-related variables concerning AI in banking. It provides practical as well as theoretical contributions by unravelling psychological and perceptual factors underlying AI trust, particularly in an emerging economy. The insights give bank managers and policymakers a conscious starting point for promoting responsible and inclusive AI adoption.

State-owned enterprises (SOEs) are pivotal to Thailand’s economy, contributing significantly to GDP, public service delivery, and national competitiveness. However, SOEs face mounting challenges from global economic uncertainty, digital disruption, and sustainability demands. This study examines how strategic governance (SG) and green leadership (GL) shape employee agility (EA) in Thai SOEs, with dynamic capabilities (DC) and knowledge management capability (KMC) serving as mediators. Drawing on the OECD Guidelines on Corporate Governance of SOEs (2 0 2 4 ), dynamic capabilities theory, and the knowledge-based view, the study employs an explanatory sequential mixed-method design. The quantitative phase uses a survey of at least 400 employees across approximately 50 SOEs, analyzed with structural equation modeling (SEM), while the qualitative phase involves indepth interviews with 1 5 executives and HR/ESG managers. The findings are expected to demonstrate that SG and GL influence EA indirectly through DC and KMC, underscoring the importance of organizational capabilities in translating governance and leadership into workforce adaptability. The study contributes theoretically by bridging institutional-level governance and leadership with micro foundational perspectives of agility, and practically by offering strategies for policymakers and SOE leaders to foster resilience and competitiveness in volatile environments.

The issue of miscommunication in HR practices remains a critical barrier in Thailand’s higher education sector, particularly for foreign lecturers whose effectiveness is shaped by cultural, linguistic, and structural challenges. While HR departments design policies, the execution of processes such as training, performance appraisal, and feedback is largely carried out by line managers, often without formal acknowledgement in policy frameworks. This creates gaps that affect lecturer performance at task, contextual, and adaptive levels. The objectives of this research are: 1) to study the causal factors of miscommunication in the execution of HR practices affecting performance, 2) to study the influence of miscommunication in the execution of HR practices affecting performance, and 3) to develop a model of the causal factors of miscommunication in the execution of HR practices affecting performance. The researcher collected qualitative data through open-ended questionnaires and interviews, which were thematically analyzed to design the survey instrument. The quantitative phase will collect data from at least 400 foreign lecturers across Thai universities between October and December 2025, with results analyzed using Structural Equation Modelling (SEM). The anticipated outcomes will provide both academic contributions, by clarifying the role of line managers in the execution of HR processes, and professional benefits, by offering practical strategies to strengthen teaching quality, lecturer engagement, and institutional competitiveness in Thailand’s higher education sector.

Cardiovascular disease is a leading cause of global mortality and predicting this disease is a crucial challenge in clinical data analysis. Effective prevention and treatment depend on early and accurate screening of affected persons. Making informed healthcare decisions relies on accurately identifying the people at risk of severe illness. The application of Machine Learning (ML) techniques and feature selection shows great potential in the early detection of cardiac disease. Machine learning algorithms are able to uncover patterns and cardiovascular health risk factors by successfully choosing the most significant factors from large datasets. This enables the creation of precise prediction models that support early intervention and detection. Thus, it improves patient outcomes in battling cardiovascular disease. In this study, I have determined the features that are most pertinent to disease prediction using wrapper feature selection technique known as Support Vector Machine - Recursive Feature Elimination (SVM-RFE). Then a Bagging Ensemble Machine Learning Classifier is used, which combines various base classifiers including k-Nearest Neighbors, Random Forests, and Decision Trees. Based on the experimental results, this ensemble approach which combines the predictions from several models attains an accuracy of 99% which outperforms other ML algorithms and similar works. The performance of the proposed ensemble learning model is further validated using the receiver operating characteristic curve and the value of ROC-AUC is 1.00. The findings demonstrate that the proposed ensemble model is capable of accurately predicting the risk of cardiac disease.

Gastrointestinal (GI) parasitism is a major health constraint in goats, causing significant losses through poor growth, reduced productivity, and increased mortality. The present study was undertaken to assess the prevalence and seasonal distribution of GI parasites in goats of the North Nashik region, Maharashtra, India. A total of 256 fecal samples were collected over eight months, of which 197 were successfully processed using sedimentation and flotation methods, along with egg per gram (EPG) estimation. Results showed that 186 samples (94.48%) were positive for one or more GI parasites. Coccidia were the most prevalent (82.4%), followed by Taenia saginata (69.27%), Giardia (22.71%), Entamoeba histolytica (9.17%), and lower rates of Dictyocaulus filaria and Moniezia sp. (3.02% each). Seasonal incidence peaked in the monsoon (96.06%), decreased post-monsoon (92.15%), and was lowest in winter (81.67%). Kids (<6 months) showed higher infection rates (95.75%) than adults (92.29%). The findings highlight a very high burden of GI parasites in goats of North Nashik, stressing the need for strategic seasonal deworming, improved hygiene, and region-specific parasite control programs.

The rapid diffusion of information and communications technologies (ICT) in developing economies has generated unprecedented opportunities for economic growth, social inclusion, and innovation. Simultaneously, it has amplified exposure to cyber threats that can undermine national development agendas, jeopardize critical infrastructure, and erode public trust. This paper examines the use and necessity of cyber-security knowledge among higher-education students in developing countries. Drawing on a comprehensive literature review, policy analyses, and empirical data from recent regional surveys, the study argues that embedding robust cyber-security curricula at the undergraduate and graduate levels is not a luxury but a strategic imperative. The paper outlines the unique risk profile of developing economies, identifies gaps in current educational provision, and proposes a multi-layered framework for integrating cyber-security education into existing programmes. Recommendations include competency-based curricula, industry-university partnerships, capacity-building for faculty, and policy support for accreditation and funding. The findings underscore that equipping the next generation of professionals with cyber-security competencies can catalyse digital transformation while safeguarding socio-economic development.