Copyright © 2026 Authors retain the copyright of this article. This article 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.

@article{204673,
        author = {Sarthak Durgesh Marathe and Achuth Abhay and Harshada Namdev Godse and Prajkta Puranik},
        title = {Beyond Audio to Action: Design, Implementation, and Comparative Evaluation of an Automated Meeting Intelligence System},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {1},
        pages = {3757-3770},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=204673},
        abstract = {This paper presents the full implementation, comparative technology analysis, and empirical evaluation of InsightFlow AI, a deployed Flask-based meeting intelligence platform. Our prior work [1] proposed a pipeline built on Whisper v3, PyAnnote 3.1, and BART-Large-CNN. When it came time to actually build the system, nearly every major component was substituted for a more practical alternative: AssemblyAI for speech recognition, Groq-hosted LLaMA-3.3-70B for summarization and action-item extraction [7], TextBlob augmented with custom keyword scoring for sentiment analysis [10], and Qdrant for vector-based retrievalaugmented generation (RAG) chat [8,9]. This paper documents why those changes were made, evaluates what they produced, and is direct about what fell short. Evaluation across five modules on real meeting audio yields a Word Error Rate of 12.95%, ROUGE-1 of 0.3474 [14], action-item extraction F1 of 1.0 on five controlled test cases, sentiment accuracy of 88.0%, and RAG chat accuracy of 75.0%, all running within a 94.2MB memory footprint with no GPU requirement. The technology substitution decisions are analyzed across five dimensions: latency, cost, scalability, maintainability, and multilingual support, and concrete future directions are identified including WebSocket-based live transcription, semantic embeddings, and third-party platform integration.},
        keywords = {meeting intelligence, speech recognition, large language models, retrieval-augmented generation, sentiment analysis, abstractive summarization, action item extraction, Flask, production AI systems, comparative evaluation.},
        month = {June},
        }