This week Talkdesk announced new Generative AI features which they broadly describe will make AI more responsible, accurate, and accessible in the contact center. As is my habit, I will take a step back to put this announcement in a broader context of Talkdesk’s AI efforts to date. I will then describe what is new and how it differs from the announcements and plans of other CCaaS leaders.
A Long History
Like many in the customer experience space, Talkdesk’s work to incorporate artificial intelligence into its solutions did not begin in November 2022, with the well-documented announcement of OpenAI’s ChatGPT. As early as November 2017, the company talked of using AI in a solution targeted for inside sales people.
It was in November 2018, in conjunction with Talkdesk’s OpenTalk user group meeting, that the company unveiled a broader AI plan for customer experience, under the brand Talkdesk IQ. As shown in the graphic, Talkdesk IQ was described as an AI layer that was part of the Talkdesk platform, supporting all applications, e.g., analytics, self-service and omnichannel interactions.
Just a few months later, at Enterprise Connect in March 2019, Talkdesk announced it was building a workforce management solution using AI technology. As I was quoted at the time:
“While we’ve seen several companies enter the Contact Center as a Service market with shiny new microservices-based solutions, workforce management (WFM) has, for the most part, been addressing the move to the cloud with variants of premises-based applications. It is exciting to see Talkdesk take on the challenge of throwing the proverbial Erlang C book out the window to build a true cloud-native, single platform, artificial intelligence (AI)-to-the-core, WFM application.”
Talkdesk continued to infuse AI into its solutions, as seen in the timeline shown below. Notable was AI Trainer in March 2021, which allows agents or others familiar with a company’s contact center operation to re-train machine learning models when they fail, to increase efficiency.
Gen AI Re-Invents Auto-Summarization
As described, the widely reported and discussed announcement of ChatGPT in November 2022 changed the AI discussion, notably in contact centers. The CX artificial intelligence conversation went from “maybe we’ll do a trial sometime in the future” to “when can I get it?” One of the first applications to be deployed by most contact center companies was auto-summarization.
After-contact work is one of the most crucial factors to consider when discussing contact center efficiency. Agents typically take several minutes to document an interaction once it is finished. This includes summarizing the call, categorizing the call outcome, escalating issues, or updating customer information. Talkdesk initially announced Agent Assist in March 2019 to reduce agent effort on these tasks using Talkdesk IQ.
In February 2023, Talkdesk was one of the first CCaaS vendors to announce that it was updating its Agent Assist capabilities ChatGPT. It’s first generative AI-powered feature, Automatic Summary and Disposition has now been in use by customers most of 2023.
New Gen AI Solutions
This week’s announcements show how much work Talkdesk has been doing in 2023 to incorporate Generative AI across the portfolio.
- Generative Knowledge Retrieval enhances the Talkdesk Knowledge Management solution by using trusted data stored in the knowledge base to deliver answers – rather than articles – to agents, creating precise responses that are simple and conversational.
- Training Data Augmentation automatically suggests intents, training phrases and synonyms, based on defined business needs, shaving considerable time off creating and managing AI models.
- Automatic Topic Discovery analyzes conversation data in the contact center and automatically detects trending topics, surfacing insights on key issues happening in the center.
- Gen AI Observability Dashboard is a tool designed to provide a comprehensive view of the performance of deployed AI solutions. The goal is to provide insight and control so that biases or hallucinations can be easily identified for fine-tuning.
- Process-based Virtual Agents using Generative AI dramatically speeds up the creation of new virtual agents, based on predefined business processes and dynamic personas.
- Gen AI Fine-Tuning and Simulation enables administrators to establish guardrails with prompt-based rules that control the output and expression of generative AI models, then simulate the results—without requiring any technical expertise.
The first four solutions described above are available now. The last two, Process-based Virtual Agents and Gen AI Fine Tuning and Simulation, are expected to be available later in the fourth quarter of 2023.
In a pre-briefing, Charanya Kannan, Talkdesk’s chief product and engineering officer, provided a sample use case of Process-based Virtual Agents. “Let’s say a customer is trying to order a train ticket (via virtual agent). In the persona design, you provide instructions on when to escalate to a human, what the tone of the conversation should be, how many times you should ask the customer to repeat data entered. This is something that was absolutely not possible before Gen AI.”
2023 has been an exciting year for the contact center and customer experience space, with most of the market leaders making or planning to make announcements around their expanded Gen AI capabilities. It is no surprise then that looking ahead to Enterprise Connect 2024, the working title for the contact center executive panel general session is, “Has Gen AI Improved CX Solutions and Outcomes Yet?” Stay tuned to see how we answer that question.