
Dana Whisper is an AI-powered real-time meeting assistant designed to help users better understand and participate in live conversations. The platform captures audio directly from the user’s device, generates live transcriptions, and provides contextual assistance during meetings without requiring integration with external meeting providers or video conferencing platforms. The system acts as a conversational copilot in real time. Users can ask questions about the ongoing discussion, request summaries, clarify technical concepts, or receive contextual suggestions on how to respond during the meeting. This makes Dana Whisper especially valuable for non-native English speakers, technical discussions, and fast-paced corporate environments where keeping up with the conversation can be challenging. One of the key strengths of the project is its architecture, which was designed with the support of IBM BOB and IBM AI technologies. IBM BOB played an important role in helping us structure the application as a scalable multi-agent system, enabling modular communication between services and intelligent orchestration across the platform. The application leverages several IBM technologies, including: IBM Watson Speech to Text for real-time transcription IBM Watson Natural Language Understanding for contextual analysis IBM watsonx.ai Granite models for intelligent response generation IBM watsonx Orchestrate for coordinating AI agents and workflows The platform is built using Electron, React, TypeScript, Node.js, Express, Socket.io, SQLite, and Drizzle ORM, following an offline-first approach where user data is stored locally for better privacy and performance. Dana Whisper’s architecture is divided into independent services for transcription, context management, phrase learning, and AI suggestion generation.
17 May 2026