Agentic AI Based Smart Assistant: A Multimodal Visual Question Answering System Using Fast API and GROQ Vision-Language Models

DOI

https://www.doi.org/10.59256/indjcst.20260502026

Published In

Volume 5, Issue 2 (2026)

Published Online

May-August 2026

Pages

237-240

Article Type

Original Article

Check for updates

This paper presents the design and implementation of an Agentic AI Based Smart Assistant — a multimodal web application capable of analyzing images and answering natural language queries in real time. The system integrates FastAPI as the backend web framework with the GROQ API, leveraging LLaMA-based Vision-Language Models (VLMs) to interpret both visual and textual data simultaneously. An additional Retrieval-Augmented Generation (RAG) pipeline using TF-IDF vectorization enables document-aware question answering from uploaded PDFs. The system was tested across ten functional scenarios including valid and invalid inputs, large images, concurrent requests, and API fault conditions — all passing successfully. Results demonstrate strong contextual accuracy and low-latency performance suitable for real-world applications in medical imaging, smart education, and automated inspection.