Project Year
Industry
A UAE-based client sought to develop a proof of concept (POC) for an AI-powered meeting assistant capable of recording online meetings, generating full transcriptions, and creating structured meeting notes. The goal was to build a cost-effective solution that did not rely on paid or official recording methods, such as Google Meet’s built-in recorder.
Client Problem:
The client required an automated solution to accurately capture meeting discussions and generate notes without manual intervention. However, existing solutions presented several obstacles:
Official recording features were limited, necessitating alternative approaches.
The system needed to autonomously join meetings audio from both the system and participants.
Existing transcription services struggled with multiple languages, particularly Arabic.
The client required an easy-to-use API for managing recording and transcription.
An alternative to built-in recording features was necessary.
A bot needed to join meetings automatically and capture audio.
The solution had to deliver accurate transcriptions in both English and Arabic.
Many transcription tools required MP3 input, while recordings were in video format.
The system needed to be scalable and integrate with tools like Google Calendar.
To address these challenges, we implemented Recall.ai for automated meeting participation and Intella for precise Arabic transcription.
A bot was configured to join meetings with minimal manual intervention.
The bot captured both microphone and system audio without notifying participants.
Participants received prompts to allow or deny the bot’s entry, ensuring compliance with security protocols.
The recorded file was emailed to the user once the meeting ended.
File Conversion: Since transcription tools required audio input, an API was developed to convert video recordings into MP3 format.
Transcription Processing: Built-in Recall.ai integrations with Azure/AWS but proved unreliable. Alternative transcription tools were evaluated.
Arabic Transcription & Diarization: Intella’s API was integrated to provide accurate Arabic transcriptions with speaker identification.
Automated scheduling and recording triggers were added.
Azure’s diarization service was integrated for better speaker differentiation.
The solution was refined to support multiple concurrent meetings.
By leveraging Recall.ai and Intella, we successfully developed a POC that addressed the client’s needs. The project evolved into a full-fledged product with enhanced integration and scalability.
Implementing NLP models for automatic meeting summaries.
Enhancing latency to provide live captions during meetings.
Strengthening data privacy features for enterprise adoption.
This case study highlights how AI-driven solutions can streamline meeting workflows and enhance productivity for businesses operating in multilingual environments.
+ 1 737 328 6762
© Copyright 2025 Qavi Tech
UAE
Germany
Saudi Arabia
New Zealand
Australia
Pakistan
