Our founder contributed to Read.ai by developing the meeting note module, dashboard interfaces, email templates, and creating bots to record meetings. Read.ai is an AI-powered meeting assistant that provides real-time transcription, smart summaries, and engagement analytics across various video conferencing platforms.
Manual note-taking during meetings was time-consuming and often led to missed information, affecting productivity.
Our founder was part of the development team, contributing significantly to the success of the project by working on the meeting note module, enabling automatic transcription and summarization of meetings.
Meeting note module development
Dashboard interface creation
Email template design
Meeting recording bot development
Real-time transcription
Smart meeting summaries
Engagement analytics integration
Designing a robust module that could accurately capture, organize, and present meeting notes in a useful format for users.
Creating intuitive dashboard interfaces that could present complex meeting data and analytics in an accessible way.
Developing dynamic email templates that could effectively summarize meeting content while maintaining personalization.
Building reliable bots that could join and record meetings across various video conferencing platforms without disruption.
Developed a modular note-taking architecture that could process speech input, identify key points, and organize content into structured, searchable notes.
Implemented an iterative design process with frequent user testing to create dashboard interfaces that presented complex meeting data in an intuitive, actionable format.
Created a sophisticated template engine that could dynamically generate personalized email summaries based on meeting content, participants, and action items.
Built a unified bot framework that could reliably join and record meetings across Zoom, Teams, Google Meet, and other platforms while maintaining high audio quality.
Implemented advanced NLP techniques to improve transcription accuracy, identify speakers, and extract key insights from meeting conversations.
Read.ai is an AI meeting assistant that joins Zoom, Microsoft Teams and Google Meet calls as a notetaker bot, transcribes in real time, and produces summaries, action items, engagement analytics and emailed recaps. The platform runs as cloud-native services on AWS: a fleet of recording bots streams audio into a Python-based speech and NLP pipeline, Node.js services persist results in PostgreSQL and object storage, and a React dashboard surfaces them. Separating capture, processing and presentation let each tier scale with meeting volume independently.
A React web dashboard, mobile apps and a browser extension for reviewing meetings, plus calendar connections that decide which meetings the bot attends.
Node.js APIs authenticate users and workspaces, serve meeting data to the dashboard, and receive callbacks from conferencing platforms.
A bot orchestrator schedules and runs recording bots per meeting; a notes service assembles transcripts into structured notes, summaries and action items; a template engine renders recap emails.
PostgreSQL stores users, meetings, transcripts and metrics; object storage keeps audio and video; queues hand recordings to the processing pipeline.
A Python speech and NLP pipeline performs real-time transcription, speaker identification, summarisation and sentiment scoring, with outputs pushed to Slack, email and webhooks.
Autoscaled AWS compute for bursty bot and transcription workloads, with the bot fleet isolated from user-facing APIs.
A user connects their calendar; the bot orchestrator schedules a notetaker for each meeting the user has opted in.
At start time a recording bot joins the Zoom, Teams or Meet call, captures audio and participant events, and streams them to the pipeline.
The Python speech service transcribes in real time and tags speakers; engagement and sentiment scores update live in the dashboard.
When the meeting ends, the recording lands in object storage and a job is queued for summarisation and action-item extraction.
The notes service stores the structured transcript, summary and metrics in PostgreSQL and indexes the transcript for search.
The template engine renders a personalised recap email and the integrations service posts it to Slack, email and configured webhooks.
Joining calls as a participant works uniformly across Zoom, Teams and Meet without per-platform recording APIs; running bots as disposable workers isolates failures to a single meeting.
Live metrics need low-latency speech-to-text, but high-quality summaries benefit from the full transcript; splitting the pipeline gave both without blocking the live view.
Keeping notes, summaries and action items as a separate module consuming the transcript let the team improve summarisation independently of capture and transcription.
Recaps vary by participants, content and action items; a data-driven template engine produced personalised emails without hand-coding each variation.
“The meeting note module developed by the team has been a game-changer for our platform, significantly enhancing our transcription and summarization capabilities.”
We ship MVPs, AI agents, and production systems in 8 weeks. Bring the rough spec — we'll scope the build inside a week with a fixed timeline and quote.
