Assisting Agents with AI
Here at XAPP, we focus on empowering users by delivering AI solutions for search and conversational experiences. Historically, we’ve defined the user as the customer, prospect, etc.
What about employees, or agents in your contact center taking calls? XAPP AI provides a solution to aid your agents, and improve satisfaction while reducing handle time.
Contact Center Agents have a demanding role for the following reasons:
Training & Domain Knowledge – Don’t you wish you could clone your best agent? You know the one, they have experience, a deep knowledge of the products and services, plus..they know the procedures of the organization. These employees are gold, but you’re lucky to have a single employee like that on the floor during a shift.
Handle Time – At the moment, contact centers are seeing massive increases in call traffic, queue times are increasing with no clear path to resolution. You know the drill, ‘an agent will be with you in [15] minutes’. In a world where people expect immediate resolutions to their requests 24/7, contact centers are desperate for solutions.
Productivity & Scale – Contact Centers are viewed as ‘cost centers’ in organizations, so the aim is to reduce cost and run the unit efficiently. That has led to a common scenario: antiquated equipment, entry-level staff with high turnover rates, and a poor experience for the caller. Agents aren’t getting the tools and support they need.
Live Agent Assist
XAPP AI is an AWS Partner with Conversational AI Competency . We collaborated with Victor Rojo (Tech Lead, Conversational AI Competency Program for AWS) to help leverage AWS’ agent assist solution using AI to solve the contact center’s challenges such as the ones laid out above. In the case of agent assist, you can augment the agent’s capabilities and reduce call handle time by transcribing the call and delivering helpful answers and scripts to improve the quality of the interactions.
The AWS agent assist solution built by Bob Strahan, Babu Srinivasan, Chris Lott, Sagar Khasnis, Kishore Dhamodaran, Oliver Atoa, and Court Schuett provides the source code, and Cloudformation scripts to modify and run the entire solution on your AWS account.
XAPP AI modified the original solution from the AWS team to address some specific challenges for a customer. They wanted a way to streamline an agent’s workflow through helpful hints, messages, and script suggestions. Then, let them click links displayed on screen to fast-forward directly into a customer’s appointment.
All Pro Plumbing
Offering plumbing services in the Tampa Florida area, All Pro Plumbing (All Pro) has seen a massive influx of traffic to their website and contact center. A large portion of each query included looking up customer information and scheduling appointments.
First, All Pro wanted to offer self-service options on their website to prospects and customers. XAPP created a conversational self-service application that uses Amazon Kendra, an intelligent search service, and our pre-built industry FAQs for plumbing services.
Then, we used our open source SDK to integrate into All Pro’s backend CRM. That integration included looking up customer accounts for authentication, providing scheduling services like booking appointments for lead capture, and rescheduling service calls for existing customers. With this effort, new transactional flows like booking and appointment, or looking up and account became self-service capabilities.
Building on that foundation, we defined a new experience for agents on our platform. We built a ‘Live Agent Assist’ channel to transcribe the calls and based on the intent of the call, an Amazon Lex bot would facilitate common agent workflows. If the bot wasn’t able to match a top-level intent, it would use Amazon Kendra to search All Pro’s knowledge base, and XAPP pre-loaded vertical industry FAQs.
Starting with Kendra, and our pre-built industry FAQs for plumbing services, we created a default experience to answer questions for visitors to their website. Then, we used our open source SDK to integrate into All pro’s backend CRM. New transactional flows like booking an appointment, or looking up an account became self-service capabilities. Building on that foundation, we were able to define a custom channel on our platform.
In this case, we built a ‘Live Agent Assist’ channel to facilitate common workflows defined by Intents through our Lex model. Then, feed transcriptions that don’t match a top level intents, or FAQs to a Kendra instance that will supplement the agents knowledge with information from the website and pre-loaded vertical industry FAQs.
In the video demonstration below, we show a comparison of what the experience is like with and without AWS AI services to assist in customer identification, scheduling, and question answering.
