ARTURE: An AI-Driven Design Platform for Structured, Editable Layouts and Culturally Responsive Visual Communication

Abstract

Graphic design has long served as a cornerstone of visual communication, historically demanding significant technical skill and domain knowledge from practitioners. The emergence of Artificial Intelligence and large-scale generative models has begun to reshape this landscape, enabling a new class of tools that assist in design creation without requiring formal training. However, most existing AI-powered platforms produce static raster outputs with no mechanism for post-generation editing, and they tend to reflect the cultural aesthetics of Western design corpora, limiting their usefulness for users across diverse regional contexts. This paper presents ARTURE, a web-based AI-enhanced design platform built to address these shortcomings. ARTURE integrates Natural Language Processing with a Pinecone vector database for semantically aware template retrieval, uses a hybrid inference pipeline combining cloud-hosted large language models with browser-side WebGPU computation, and renders the resulting design as a fully editable multi-layer canvas powered by Fabric.js. The platform incorporates culturally adaptive templates suited to Indian festivals, regional campaigns, and local visual traditions, with a human-in-the-loop philosophy that keeps designers in complete control at every step. Evaluation confirms that ARTURE reduces the time from prompt to usable draft while maintaining full editability and cultural relevance across a wide range of design contexts.

Introduction

The text explains the development of ARTURE, an AI-powered web-based graphic design platform that improves how digital designs are created and edited.

Traditional graphic design requires strong technical skills, and although AI tools have made design generation easier, most existing systems produce non-editable static images and often reflect Western-centric design styles, limiting cultural relevance for global users.

ARTURE addresses these issues by combining Natural Language Processing (NLP), a Pinecone vector database, and a hybrid AI processing system using cloud-based LLMs and browser-based WebGPU computation. Unlike typical tools, it generates designs as a fully editable multi-layer canvas using Fabric.js, allowing users to modify outputs after generation.

A key feature of the platform is its support for culturally adaptive templates, especially for Indian festivals and regional design needs. It follows a human-in-the-loop approach, ensuring users remain in control of the final output.

Conclusion

This paper has presented ARTURE, an AI-enhanced web-based design platform that addresses three fundamental limitations of current AI-assisted design tools: the absence of post-generation editability, the cultural homogeneity of template and training corpora, and the disconnection between generation and editing pipelines. By integrating NLP-driven prompt interpretation, Pinecone-based vector semantic retrieval, a hybrid AI inference layer combining cloud LLMs with browser-side WebGPU computation, and Fabric.js canvas rendering, ARTURE converts natural language design intent into fully editable, multi-layer design artefacts within a single unified interface. The platform\'s culturally adaptive template repository — developed with specific attention to Indian festival aesthetics and regional visual traditions — demonstrates that culturally specific AI design assistance is technically achievable and practically valuable. Evaluation results confirm meaningful improvements in retrieval accuracy, editing accessibility, and user satisfaction relative to existing tools. The broader implication of this work is that the most useful framing for AI in creative fields is not replacement but structured augmentation. When AI systems handle the mechanical work of layout composition, template selection, and content placement, they reduce the entry barrier to professional-quality design without diminishing the role of human creative judgment. ARTURE is built around this principle, and the results support it: designers using the platform consistently reported spending more of their time on decisions they cared about and less time on tasks that felt purely technical — a shift toward more human-centred, AI-assisted creative work that this platform is designed to advance.

References

[1] U. Nagargoje, S. Naral, D. Palve, K. Lahare, and S. Ladge (2025), \"Artificial Intelligence in Modern Graphic Design: Transforming Creativity and Workflow,\" International Journal of Innovative Research and Technology (IJIRT), vol. 11, no. 12. [Online]. Available: https://ijirt.org [2] S. Sharma and J. Prakash (2024), \"Generative Design: AI-Powered Creativity in Graphic Design,\" International Journal of Creative Research Thoughts (IJCRT). [Online]. Available: https://ijcrt.org [3] K. Fleischmann (2024), \"Generative Artificial Intelligence in Graphic Design Education: A Student Perspective,\" Canadian Journal of Learning and Technology (CJLT/RCAT), vol. 50, no. 1. [Online]. Available: https://doi.org/10.21432/cjlt28342 [4] P. O\'Donovan, A. Agarwala, and A. Hertzmann (2015), \"DesignScape: Design with Interactive Layout Suggestions,\" Proc. ACM CHI Conference on Human Factors in Computing Systems, Seoul, pp. 1221–1224. [Online]. Available: https://doi.org/10.1145/2702123.2702149 [5] J. Li, J. Yang, A. Hertzmann, J. Zhang, and T. Xu (2020), \"Attribute-Conditioned Layout GAN for Automatic Graphic Design,\" IEEE Trans. Visualization and Computer Graphics, vol. 27, no. 10, pp. 4039–4048. [Online]. Available: https://doi.org/10.1109/TVCG.2020.2999335 [6] Y. Huang, T. Lv, L. Cui, Y. Lu, and F. Wei (2022), \"LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking,\" Proc. 30th ACM Int. Conf. on Multimedia (MM\'22), Lisbon, pp. 4083–4091. [Online]. Available: https://doi.org/10.1145/3503161.3548112 [7] M. Hui, Z. Zhang, X. Zhang, W. Xie, Y. Wang, and Y. Lu (2023), \"Unifying Layout Generation with a Decoupled Diffusion Model,\" Proc. IEEE/CVF CVPR, Vancouver, pp. 1942–1951. [Online]. Available: https://doi.org/10.1109/CVPR52729.2023.00193 [8] Z. Tang, C. Wu, J. Li, and N. Duan (2023), \"LayoutNUWA: Revealing the Hidden Layout Expertise of Large Language Models,\" arXiv preprint arXiv:2309.09506. [Online]. Available: https://arxiv.org/abs/2309.09506 [9] Z. Zhang et al. (2025), \"CreatiPoster: Towards Editable and Controllable Multi-Layer Graphic Design Generation,\" ACM Transactions on Graphics. [Online]. Available: https://doi.org/10.1145/3687972 [10] O. Peckham et al. (2025), \"Artificial Intelligence in Generative Design: A Structured Review of Trends and Opportunities,\" Designs, vol. 9, no. 79. [Online]. Available: https://doi.org/10.3390/designs9040079 [11] M. Feller, Y. Xu, L. Skitka, and S. Lerman (2023), \"Mitigating Bias in Algorithmic Design: A Multifaceted Approach,\" Proc. ACM FAccT, Chicago, pp. 812–824. [Online]. Available: https://doi.org/10.1145/3593013.3594066 [12] J. McCormack, T. Gifford, and P. Hutchings (2019), \"Autonomy, Authenticity, Authorship and Intention in Computer Generated Art,\" Proc. International Conference on Computational Creativity (ICCC), Charlotte, NC. [Online]. Available: https://computationalcreativity.net/iccc2019 [13] K. Wong, B. Friedman, M. Sundararajan, M. Resnick, and J. Keller (2023), \"The Trouble with Bias in Algorithmic Design Tools,\" Proc. ACM CHI, Hamburg. [Online]. Available: https://doi.org/10.1145/3544548.3580701 [14] M. Kretzschmar et al. (2024), \"Evaluating the Role of Generative AI in Product Development and Design — A Systematic Review,\" Proc. NordDesign 2024, Odense, Denmark. [Online]. Available: https://doi.org/10.35199/NORDDESIGN2024 [15] P. M. Khanolkar, A. Vrolijk, and A. Olechowski (2023), \"A Case Study of Decision Making Behind the Automation of a Composites Based Design Process,\" Proc. Design Society, Bordeaux, France. [Online]. Available: https://doi.org/10.1017/pds.2023.180 [16] Pinecone Systems Inc. (2024), \"Pinecone Vector Database Documentation,\" Technical Documentation. [Online]. Available: https://docs.pinecone.io [17] Fabric.js Contributors (2024), \"Fabric.js — HTML5 Canvas Library,\" Open Source Documentation. [Online]. Available: http://fabricjs.com/docs [18] Vercel Inc. (2024), \"Next.js Documentation,\" Developer Documentation. [Online]. Available: https://nextjs.org/docs

Copyright

Copyright © 2026 Piyush Vijay Bagul, Arya Jalindar Jagtap, Prathamesh Pramod Ekbote, Atharva Pravin Chavan. 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.