AI customer experience examples add to CX playbook
Large language models such as GPT-3 are paving the way for a step-change in the adoption of AI by firms, particularly when it comes to the way that companies interact with their customers. With the excitement of advanced chatbots – ChatGPT being the current leader of the pack – comes concerns. If AI can answer client queries, what does that mean for contact center staff and other workers? It’s a valid question, but AI customer experience success stories point to a brighter and more interesting future. And with 84% of C-suite executives believing that scaling AI puts firms on a path to achieving their growth objectives, based on Accenture findings, the direction of travel is clear.
Investing in customer engagement
The research report, authored by Accenture’s Athena Reilly, Joe Depa, and Greg Douglass, highlights that C-suite executives see AI as an enabler rather than a threat. However, the rise of AI customer experience solutions – focusing on the CX sector as an early adopter of machine learning and other digital tools to accelerate productivity – does bring a sense of urgency. Putting off investment and taking a ‘wait and see’ approach could prove costly if competitors are proactive in embracing solutions enabled by foundation models such as GPT-3. API access is making advances in AI available more widely, and leading firms are wasting no time in discovering what the latest digital breakthroughs can bring to the CX playbook.
In March, enterprise conversational AI platform developer Aivo announced that it was testing ChatGPT integration with its in-house conversational engine. Aivo’s product suite gives clients the ability to automate conversational journeys across a variety of channels. For example, Banco Comafi – an Aivo client – started with a web chat interface and then added a WhatsApp integration, giving its customers the ability to interact with the South American bank 24/7. Increasing in popularity year-on-year, the AI customer experience was soon generating over 150,000 interactions per month (with a response time of less than 60 seconds). And the development team reports high customer satisfaction scores, backing up the growth in conversational AI.
Like many others, Aivo points out that ChatGPT may not be entirely accurate when generating responses. ChatGPT’s creator, OpenAI, cautions that its advanced chatbot, which has taken the world by storm – can ‘occasionally’ produce incorrect answers and has limited knowledge of world events after 2021. And when Microsoft demonstrated its updated Bing Search featuring ChatGPT, observers noted factual errors in the output. But there are still gains to be made under the hood in terms of producing more capable AI customer experience platforms, as Aivo has identified in its ChatGPT analysis.
The capacity of large language models to generate synthetic conversational data can help firms such as Aivo and other platform providers to accelerate the development of their in-house AI conversational engines. “This allows us to shorten significantly the time to improve our models by reusing the knowledge acquired by the foundational models in a selective and controlled way,” writes Sergio Soage – a Machine Learning Engineer at Aivo – in a recent blog post on the topic of ChatGPT: Analysis and implications for conversational technology.
Service bot parameters
OpenAI’s playground tool highlights the range of parameters that can be adjusted through API calls. For example, so-called temperature values can be lowered to make responses more precise or increased to encourage OpenAI’s advanced chatbot to generate chattier, or more creative, answers. In this way, developers can mimic a range of styles to simulate how people may query company chatbots to put AI customer experience platform upgrades to the test ahead of launch.
Text-based interactions are valuable for developing more engaging and efficient CX solutions, but there’s much to be gained from voice data too. “Call recordings are hugely valuable to customer service teams,” highlights Aircall, a provider of cloud-based call centers and phone systems for businesses. “Call recordings can be great real-world examples of what good and bad calls sound like, pinpointing the exact elements to target and avoid and where extra training is needed.”
Hosted in the cloud, recordings can be integrated across a range of enterprise systems – for example, Aircall’s solution can be dropped into HubSpot, and the call data added to client records. The productivity gains of having client call data readily accessible across product teams is already a big win, but adding AI and transcription insights on top of the original voice tracks brings more. Aircall’s app ecosystem includes a variety of AI and transcription plug-ins from providers such as Avoma, Convin, and many more. Solutions efficiently convert speech to text and make it possible to jump to specific keywords in the conversation. Tools can also give sales teams AI-generated notes for key topics, and make sure that action items get prioritized.
These are just a few examples highlighting why AI and CX is becoming such a powerful combination across a wide range of industry sectors – fintech, banking, retail, e-commerce, telecoms, and more. And to emphasize how important AI customer experience and other productivity examples could be, it’s worth returning to the Accenture analysis to spotlight other growth areas. Across all industries, 75% of executives surveyed believed that they risked going out of business in five years if they didn’t scale AI. In travel and transport, the proportion was 77%, and for utilities firms, four-fifths of respondents put AI on the must-have list.
20 March 2023