Supporting a Fortune 100 Retailer with Real-Time AI Voice Assistance on Salesforce 

‍Client Overview

We work with a Fortune 100 retail organization that distributes to management teams, property operators, and purchasing departments in the USA.

The call center of the above company receives a number of phone calls on a daily basis from customers.

With Salesforce Service Cloud at the center of their operations, the client needed a way to back agents up during live calls without pulling them out of the conversation.

The Challenges: No In-Call Support, Gaps in Recall, Disjointed Case Creation

Voice support placed a different kind of pressure on agents. Calls move in real time with no room to pause and look things up. Which led to some issues showing up often, like:

• Agents Lacked Real-Time Guidance During Complex Calls: When a complicated question was presented, the agents would place the caller on hold or work from memory, causing a slowdown and inconsistent responses among the team members.
• Difficult to Track Critical Information During Long Calls: No record existed of what was being said during the call. This left the agents with no means to revisit their earlier dialogue points. Meaning that points raised earlier in the conversation were likely overlooked and needed to be reiterated.
  
• Case Creation Sat Entirely With the Agent: Unlike other channels, voice interactions did not automatically generate a case. Agents logged everything manually after each call from memory, leading to incomplete records and post-call admin that built up quickly.

Our Intelligent Solution

We placed Thunai inside the client's Salesforce Service Cloud environment as a native LWC widget connected through the Service Cloud CTI. Thunai ran alongside every live call through two dedicated panels.

• Real-Time Reply Suggestions via Voice Live Agent Assist: As the customer speaks, Thunai reads the conversation and surfaces reply suggestions directly in the agent's Salesforce panel. Agents draw on these mid-call without putting the customer on hold. In doing this, agents always have a policy-backed response within reach, regardless of what comes up on the call.
  
• Live Call Transcription via the Voice Transcript Panel: The Voice Transcript panel captures the full conversation as it plays out. Agents can refer back to earlier parts without asking the customer to repeat themselves. With this, details raised at the start of a call stay visible all the way through to close.
  
• Documentation-Backed Suggestions via Thunai Brain: Every suggestion is drawn from the client's SOPs and internal procedures in Thunai Brain. Agents are not working from memory mid-call. But on the other hand, where a query falls outside standard documentation, the agent is flagged to take the appropriate next step.
  
• Agent-Led Case Creation After the Call: Voice interactions do not automatically generate a case in Salesforce. Agents create the case manually once the call closes, drawing from the live transcript and suggestion log as a reference for an accurate record.

Conclusion

By combining real-time transcription with on-call assistance, Thunai gave agents the backing they needed on live calls without asking them to break from the conversation or work across multiple screens.

Suggestions grounded in actual company documentation mean agents are no longer relying on memory alone when complex queries come in. In doing this, responses across the team have become more consistent. 

Thunai enabled agents to handle customer conversations with greater confidence, consistency, and speed while reducing the operational burden associated with post-call documentation.