VisoryBI: A Template-Driven Offline Business Intelligence Platform with Integrated Statistical Analytics and Natural Language Querying

Abstract

Business Intelligence platforms have become indispensable in modern organizational workflows, yet most production-grade tools demand persistent internet access, vendor-managed servers, and considerable technical proficiency. These constraints effectively exclude a wide range of users, from small-business analysts to data teams operating in air-gapped environments. VisoryBI was built specifically to remove those barriers. It is an offline-capable, browser-native dashboard builder implemented as a Progressive Web Application that performs all computation, storage, and rendering on the client device without any backend infrastructure; after the initial load, the application operates without any network connection. The platform ships with more than seventeen chart types, sixteen domain-specific dashboard templates equipped with intelligent field-mapping, an integrated statistical engine performing Z-Score and Interquartile Range anomaly detection alongside linear-regression forecasting, and a client-side Natural Language Processing interface that lets users query their datasets in plain English. Persistence is handled through a two-tier storage model pairing Zustand-managed localStorage for lightweight state with IndexedDB for large datasets. Export pipelines support PDF, PNG, PowerPoint, and CSV output. A three-tier role-based access control system governs who may create, modify, or view content. Independent end-to-end testing across all modules recorded an overall working efficiency of approximately ninety-seven percent.

Introduction

Most existing BI systems rely on cloud infrastructure, require technical expertise, and restrict advanced analytics in free versions. This creates barriers for users in low-connectivity environments, non-technical users, and organizations with strict data privacy needs. VisoryBI addresses these issues by offering a zero-installation, fully offline Progressive Web Application (PWA) that runs entirely in the browser without a backend or internet connection after loading.

The system allows users to upload datasets (CSV/Excel), automatically detects data types, and uses a smart field-mapping engine to instantly generate dashboards. It provides 17 visualization types, 16 dashboard templates, drag-and-drop dashboard building, and real-time filtering.

It also includes an in-browser analytics engine featuring anomaly detection (Z-Score and IQR), trend forecasting using regression, and statistical summaries. A natural language query system (Ask Data) lets users ask questions in plain English, which are converted into visual outputs without external APIs.

VisoryBI supports offline storage using localStorage and IndexedDB, role-based access control (Admin, Editor, Viewer), dashboard versioning, and export options such as PDF, PNG, PowerPoint, and CSV. All processing, including NLP and analytics, happens locally in the browser.

Research cited in the paper shows improvements in dashboard design, layout optimization, and visual attention, which are incorporated into VisoryBI’s design principles, such as KPI placement and structured templates.

Evaluation results show the system achieves about 97% functional efficiency and outperforms other BI tools in offline capability, automation, and accessibility while remaining free and lightweight.

Conclusion

This paper presented VisoryBI, an offline-capable browser-based Business Intelligence dashboard builder that addresses the connectivity, expertise, and analytics capability gaps that limit adoption of existing BI platforms. The system provides seventeen visualization types, sixteen domain-specific templates with automated field mapping, a two-tier client-side storage architecture, Z-Score and IQR anomaly detection, linear-regression forecasting, a client-side NLP query interface, PowerPoint export, a fullscreen presentation mode, dashboard versioning, and a three-tier role-based access control system. End-to-end evaluation recorded an overall working efficiency of approximately ninety-seven percent. The design of VisoryBI was informed throughout by recent research in dashboard visualization. The NLP query interface and drill-down navigation implement the turn-taking and repair modes identified by Setlur et al. [1]. The template architecture encodes the layout rules derived by Lin et al. [2]. Default widget arrangements reflect the attention patterns established by Yang et al. [3]. The PowerPoint export pipeline applies the aspect-ratio and topology-preservation principles validated by Zeng et al. [4]. Planned future work includes upgrading the forecasting engine from linear regression to ARIMA or exponential smoothing models, extending the NLP engine with a transformer-based intent parser, adding chart annotations including reference lines and goal-line markers, implementing bcrypt-based password hashing for production hardening, and introducing conditional formatting rules that automatically highlight KPI cards in green, amber, or red based on user-configured target thresholds.

References

[1] V. Setlur, M. Correll, A. Satyanarayan, and M. Tory, “Heuristics for Supporting Cooperative Dashboard Design,” IEEE Trans. Vis. Comput. Graph., vol. 30, no. 1, pp. 370–380, Jan. 2024. [2] Y. Lin, H. Li, A. Wu, Y. Wang, and H. Qu, “DMiner: Dashboard Design Mining and Recommendation,” IEEE Trans. Vis. Comput. Graph., vol. 30, no. 7, pp. 4108–4121, Jul. 2024. [3] M. Yang, Y. Hou, L. Li, R. Chang, and W. Zeng, “Dashboard Vision: Using Eye-Tracking to Understand and Predict Dashboard Viewing Behaviors,” IEEE Trans. Vis. Comput. Graph., vol. 31, no. 10, pp. 6930–6945, Oct. 2025. [4] W. Zeng, X. Chen, Y. Hou, L. Shao, Z. Chu, and R. Chang, “Semi-Automatic Layout Adaptation for Responsive Multiple-View Visualization Design,” IEEE Trans. Vis. Comput. Graph., vol. 30, no. 7, pp. 3798–3812, Jul. 2024. [5] S. K. Badam and N. Elmqvist, “Effects of Screen-Responsive Visualization on Data Comprehension,” Inf. Vis., vol. 20, no. 4, pp. 229–244, 2021. [6] A. Sarikaya, M. Correll, L. Bartram, M. Tory, and D. Fisher, “What Do We Talk About When We Talk About Dashboards?” IEEE Trans. Vis. Comput. Graph., vol. 25, no. 1, pp. 682–692, Jan. 2019. [7] B. Bach et al., “Dashboard Design Patterns,” IEEE Trans. Vis. Comput. Graph., vol. 29, no. 1, pp. 342–352, Jan. 2023. [8] A. Wu et al., “MultiVision: Designing Analytical Dashboards with Deep Learning Based Recommendation,” IEEE Trans. Vis. Comput. Graph., vol. 28, no. 1, pp. 162–172, Jan. 2022. [9] S. Shin, S. Chung, S. Hong, and N. Elmqvist, “A Scanner Deeply: Predicting Gaze Heatmaps on Visualizations Using Crowdsourced Eye Movement Data,” IEEE Trans. Vis. Comput. Graph., vol. 29, no. 1, pp. 396–406, Jan. 2023. [10] M. Tory, L. Bartram, B. Fiore-Gartland, and A. Crisan, “Finding Their Data Voice: Practices and Challenges of Dashboard Users,” IEEE Comput. Graph. Appl., vol. 43, no. 1, pp. 22–36, Jan./Feb. 2023.

Copyright

Copyright © 2026 H. Mohamed Hashim, S. Mohamed Thasleem, S. Abdul Thowfik, M. Naveen, Mrs. J. Rosalind Geetha. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.