GreyLabs AI bets on GenAI to tune into customer conversations and draw smart insights

The Mumbai-based speech analytics startup uses large language models to help financial services institutions understand customers’ needs and improve service based on interactions with them over calls and emails.

Thursday August 08, 2024 , 5 min Read

Conversations with customers provide a tonne of unstructured data from which valuable insights can be derived. This is where speech analytics comes into play.

Now, with advancements in technology, speech-to-text analytics has become a much easier process.

Mumbai-based startup GreyLabs AI—which offers speech analytics solutions for financial services institutions—claims to do this not just with ease but also quickly, smartly, and with “near-human-level accuracy”—thanks to the integration of generative AI (GenAI) into the field of transcription.

GreyLabs’ GenAI-powered platform analyses recordings of calls with customers, monitors the service agent’s performance, and provides insights to improve customer service and quality assurance. The tool also identifies revenue opportunities from calls and flags them for the outbound team to reach out to customers for lead generation or sales.

Regulatory compliance demands that financial institutions store all customer call recordings for at least seven years. Therefore, they sit on a huge amount of unstructured data, which could provide valuable opportunities to upsell or improve customer service.

Harshita Srivastava, Co-founder and CTPO, GreyLabs AI, explains, “A lot of calls come to call centres, but a lot of opportunity—to upsell or cross-sell—gets missed by the customer service agent.

“Imagine a customer calls a bank for a change of address. This might indicate that the customer has relocated. They may have also bought a house. Our system can understand this from the call recording and flag it as an opportunity to find out if the customer is interested in a home loan.”

Srivastava, an alumnus of IIT Kanpur, co-founded GreyLabs AI last year along with Aman Goel, an alumnus of IIT Bombay. Before founding GreyLabs, they had founded Cogno AI, an AI-powered omnichannel cloud communication platform, which catered to B2B enterprises in the BFSI domain. Cogno AI was acquired by Exotel in 2021.

Commenting on what GenAI brings to the table, Srivastava says, “Companies previously converted speech into text and relied on keyword-based solutions. Agents learned to ‘stuff’ certain keywords into their calls to achieve higher scores. Generative AI, however, evaluates sentences, phrasing, context, meaning, and semantics, making the system much more accurate and intelligent.”

She adds that, with GenAI, models are pre-trained on a broad range of common sentences, words, and phrases, thus reducing the need for base training.

How the platform works

Clients upload the customer call recordings on Greylabs’ platform after the conversation has taken place. The startup’s SaaS product converts every single call placed to a customer service agent into text using a proprietary speech-to-text engine. Then GenAI-powered LLMs (large language models) are run on top to analyse the conversation and offer accurate insights.

The insights help financial services institutions understand how helpful their customer service agents are, what kind of training they need, how satisfied their customers are, and what areas need improvement. The platform also provides detailed analytics on customer chats and emails.

This automation reduces the manual effort required for call audits, speeds up the process, and enhances accuracy, says the startup’s co-founder.

Features and use cases

GreyLabs’ platform is customisable for various parameters and use cases such as sales, collections, and customer service for banks, NBFCs, life insurance, health insurance, and broking.

Every call recording can be accurately transcribed and converted into a vector of scores across 50+ audit parameters, using proprietary BFSI-tuned GenAI technology, explains Srivastava.

“The features are quite customisable without the need for much coding–customers can do it themselves,” she adds.

Here are some use cases:

Assess customer service: Institutions can understand if the agent helped the customer properly and understood their sentiment.

Identify leakages in the sales funnel and understand if the agent has pitched the product properly: Calls are graded across parameters such as objection handling, comparison with competitors, educating the customer, politeness, patient listening, and highlighting brand value.

Understand compliance: Agents make collection calls to get customers to make payments. For instance, when a customer hasn’t paid the EMI, the agent has to explain the consequences of non-payment to them. Financial institutions can find out if the agent’s behaviour complies with the company policy or if any issue escalates.

All these actionable insights ultimately help businesses finetune their customer strategy and improve their net promoter and customer satisfaction scores, says Srivastava.

Growth and future plans

GreyLabs has about eight clients in the BFSI space—including banks, insurance companies, broking firms, and NBFCs.

The startup, which is in its early stage of development, is focused on nailing down the requirements of customers and addressing their pain points.

In June this year, the speech analytics platform raised over $1.5 million in a seed round led by Z47, formerly Matrix Partners India. The company plans to use funds on product development and go-to-market strategy.

“There is an immense amount of unstructured data generated from calls and emails across businesses that remain underutilised for decision-making. In particular, regulated industries like banks and financial services need secure and compliant solutions to extract business value from this data,” says Pranay Desai, Managing Director, Z47.

“Greylabs has a significant opportunity to drive revenue growth, enhance governance, and improve customer satisfaction in these industries by leveraging secure LLMs,” he adds.