Get to know our approach

Let us walk you through our proven process for creating voice assistants, so you know what to expect when working with us.

Talkie Our apprach to building virtual agents

1. Choose the right process to automate

We start off by working with your team to identify which processes should be automated and in what order. To assess these processes, we consider different metrics such as call volume, percentage of cases solved following a script, and the amount of creativity and empathy needed to handle a particular case. 

2. Understand goals and cases for automation

In this stage, we’re focused on gaining a deep understanding of the process we’re going to automate. We map out every task and action needed for a successful implementation by analyzing i.a. your call center data and internal procedures.

At the end of this stage, we put forward: 

  • Recommendation for the implementation scope
  • Specific metrics we’ll use to measure the virtual agent performance

“From the very beginning of the project, we work closely with our clients. When architecting the solution, we rely on their input, whether it’s participation in discovery workshops or sending over call center data. Good cooperation ensures that we share a common understanding of what it means for the virtual agent to be successful and what steps need to be taken to get there.”

Ada Adruszkiewicz, Head of Product

3. Prepare virtual agent conversation script

The script includes the core conversation flow, possible paths, and customer utterances with corresponding bot reactions. It helps us better understand what you expect from the conversation and make sure that all use cases are covered. While the script serves as a foundation, we’ll keep improving it to better fit virtual agent goals and capabilities.

4. Build the virtual agent using the Talkie platform

Our customer experience team configures your virtual agent using the Talkie platform. They play the role of customer advocates and make sure that the solution works in a way that makes the human-machine interactions as smooth and natural as possible. While the solution isn’t ready to be used by customers at this stage, we encourage you to try it out as early as possible so we can act on your feedback. 

“While our bot analysts perfect the conversation flow, our developers take care of the technical side of the project. For example, if the virtual agent needs to be integrated with tools and systems you already use, our dev team works with yours to figure out the best way to do it.”

Wojciech Przechodzeń, Head of Technology

5. Test and train the virtual agent

Our team uses a suite of automated tests as well as picks up the phone and talks to the virtual agent exactly how customers would. We also invite you and your team to join this process to ensure that the solution meets your expectations. While most of the testing and training is done at this stage, it continues as long as the virtual agent works for you.

“Our clients are sometimes worried that even though the virtual agent works great during tests, it won’t be able to deal with real customer queries. To make sure that isn’t the case, we run exhaustive tests and training before launch, as well as tailor-make the rollout schedule (e.g., the first few days AI agent operates only a few hours a day) to make the transition as smooth as possible.”

Daria Pszenna, Analyst & Conversation Designer

6. Serve real customers

When a virtual agent goes live, we closely monitor the conversations to improve the solution based on real users’ data. We analyze how many calls end with success and check for interactions that proved difficult to customers. Regular reporting helps us prioritize our work on further improvements and lets you stay on top of the project’s most important metrics.

Automate your customer service now