Unlocking the Power of Post-Call Analytics: Dezerv's Journey with Sarvam
In the fast-paced world of wealth management, where every conversation can shape a client's financial future, Dezerv has found a powerful ally in Sarvam's cutting-edge technology. This partnership is not just about transcription; it's about unlocking a treasure trove of insights that were previously hidden within the audio recordings of client calls. Let's dive into the fascinating story of how Dezerv transformed its post-call analytics and the profound impact it had on the firm's growth and client relationships.
The Wealth Management Conundrum
Dezerv, a forward-thinking wealth management firm, recognized a critical challenge in the Indian market. As wealth grows, so does the need for expert management, but the expertise itself was in short supply. Dezerv's innovative solution was to design customized investment portfolios, offered through SEBI-registered Portfolio Management Services, catering to clients seeking exposure beyond traditional markets. With a team of 650+ professionals across five cities, they were on a mission to bridge the gap between wealth and expertise.
At the heart of this model were the relationship managers (RMs), who played a pivotal role in building long-term trust with clients. These RMs conducted video conferencing calls, explaining portfolio decisions and providing guidance through volatile markets. However, the true value of these conversations lay in the insights they offered, yet they were locked within audio recordings, inaccessible for in-depth analysis.
The Transcription Challenge
The transcription process was a complex task, especially given the unique linguistic landscape of urban India. Clients and RMs seamlessly switched between English, Hindi, and regional languages, often peppered with technical fund names and numbers. Accuracy was paramount, particularly when dealing with financial figures. Dezerv's previous outsourcing approach had its limitations, as they couldn't audit quality, and the transcripts, especially for Indic languages, fell short of their needs.
Enter Sarvam: A Game-Changer
Dezerv's search for a solution led them to Sarvam, a Speech-to-Text platform, after a rigorous benchmarking process. The team evaluated multiple platforms using open-source datasets, including AI4Bharat's Svarah, measuring Word Error Rate (WER) and Character Error Rate (CER) across various languages. Sarvam emerged as the clear winner, offering three key advantages:
- Accuracy on Real Indian Speech: Sarvam's models were trained on a vast dataset of Indian audio, capturing code-mixed conversations, regional accents, and telephony recordings. This made it perfectly suited to Dezerv's calls, ensuring high accuracy in transcription.
- Diarization Built-in: The platform's speaker diarization feature automatically separated different speakers, a crucial aspect in wealth management. Knowing who said what was as important as understanding the content.
- Scale Without Operational Complexity: Dezerv could submit large volumes of conversations asynchronously, receiving structured transcripts without the need to build or manage transcription systems, simplifying their operations.
The Solution in Action
Dezerv integrated Sarvam through its SDK, and the results were transformative. They implemented three production use cases:
- Transcription of Video Meetings: Sarvam's ability to handle video conferencing calls seamlessly provided structured transcripts, capturing speaker identities and conversation details.
- Transcription of Phone Calls: The platform's efficiency in transcribing phone calls ensured that every conversation was documented accurately.
- Live Note-Taking for RMs: Relationship managers could now take notes during client conversations, providing context and clarity for future interactions.
With client consent, recordings were processed via Sarvam's Batch API with speaker diarization, converting each conversation into a structured transcript. This transcript captured not just the words but also the speakers and timing, creating a searchable record of the client relationship. RMs could instantly revisit discussions, understand client needs, and improve their advice.
The Impact and Beyond
The impact of this partnership was immediate and significant. Dezerv processed close to 3.6 lakh minutes of speech through Sarvam last month, including video meetings and phone calls. The transcription quality improved, with a 20% reduction in Word Error Rate. In-house pipeline management provided direct visibility into performance and quality, enabling Dezerv to bring its insights module in-house.
Looking ahead, Dezerv's vision is to build a wealth management firm that is expert-led, technology-first, and accountable for outcomes. Learning from every client conversation is integral to this vision. As conversation volumes grow and Sarvam advances its speech models for Indian languages, the value of institutional knowledge will compound, shaping the future of wealth management in India.
In my opinion, Dezerv's journey with Sarvam is a testament to the power of technology in transforming industries. It's not just about transcription; it's about unlocking the wisdom within every conversation, enabling wealth managers to provide better advice at scale. As the wealth management landscape evolves, partnerships like this will be crucial in driving innovation and client satisfaction.
