Artificial Intelligence (AI), the simulation of human intelligence by machines, has been around for longer than some might think: The actual beginnings of this field lie in the 1950s, at Dartmouth College in New Hampshire. Over the decades, the excitement has alternately waxed and waned. What catapulted AI to the absolute highest hype levels, however, was the release of ChatGPT by OpenAI in November 2022. ChatGPT — which stands for Chat Generative Pre-trained Transformer — is a Large Language Model (LLM) that unleashed a new era of usability of AI.

The initial enthusiasm was immense. Speculation about all the problems AI will solve, and what problems it
will create, abounded. Almost a year later, the buzz has died down a little, but governments all over the world are considering legislation as to how AI can and should be used responsibly. For private use, AI capabilities are used for entertainment and, not least because of cost, will probably remain so for a while to come. Businesses, meanwhile, are ready to hit the ground running. It is no surprise that Microsoft has embedded a variety of AI-piloted features in its products.

Although it gets the most attention due to its revolutionary capabilities, ChatGPT is not all that AI can do. There are many different layers, and their targeted and pragmatic use can help organisations in their digital transformation.

Machine learning algorithms are not programmed. Instead, they “learn” from user input to solve specific problems, continuously advancing these learning capabilities. They can, for example, extract relevant data from customer orders sent in a variety of formats: free-text emails, PDFs, images and even, for those still stuck in the 90s, faxes. Sometimes, however, misidentifications occur, and values are not correctly read. This is where human intervention is still required to rectify these errors. The machine learning AI will then remember the correction, and, with the help of historical data analysis, from there on out rapidly predict increasingly correct outcomes.

Deep learning, a subcategory of machine learning, can execute even more advanced tasks. Relying on artificial neural networks that mimic the human brain and on enormous amounts of validated data, deep learning extracts information from multiple data layers, yielding even higher accuracy and optimisation. Ideal business uses for deep learning technology can be found, for example, in email triage for shared inboxes, both on the supplier and customer sides. Questions, invoices and purchase orders can all be filtered out and forwarded to the correct recipient. For the Finance department, deep learning can even pinpoint small changes in payment behaviours, order dynamics and credit ratings. These indicators can have a significant impact on working capital requirements, and thus, the earlier they are detected, the better the business can react accordingly.

And then, of course, there is ChatGPT. In addition to the fun moments it can provide when entering a query,
for example, to generate an article about AI business use-cases in the voice of Yoda, it does have some real-life applications. Customers who email a business with questions such as “Where is my order?” or “When will my reimbursement arrive?” are all living different experiences. While sentiment analysis can identify when a customer might be getting frustrated, ChatGPT can help in generating an appropriate response. This refers to both tone and content: pulling the correct information from the different systems and then suggesting a fitting answer. The emphasis here lies on “suggesting.” A human should always remain in control and make necessary adjustments, rather than relying on ChatGPT for fully automated responses. The potential for errors is just too great otherwise.

The hoped-for result would be an efficient workplace that cuts out repetitive and meaningless tasks by using AI, and brings to light hidden financial details and business indicators. These tools need to remain assistive rather than replacing humans, since no matter how intelligent AI is, the human touch just cannot be replaced. Properly used, with an awareness of potential risks, AI technologies should be used to increase well-being, strengthen our democracies, and improve the shared information environment.

Esker UK

Unlocking Positive-Sum Growth with AI-Driven Business Solutions for P2P & O2C Cycles

Read more insights from Esker UK

Built into Esker’s Customer Enquiry Management solution, ChatGPT provides Customer Service (CS) professionals with an additional resource to efficiently handle customer enquiries such as order status, availability, pricing requests and product information questions. This added element of AI assistance enables CS teams to answer customer requests on time, increase customer satisfaction, and focus on more impactful and proactive outbound efforts.

Shared inboxes allow multiple CSRs to view and reply to emails, but retrieving accurate data like order, pricing and shipping data as well as previous conversations from a variety of systems (ERP, email, CRM, chat) can add hours of work and is error-prone. With AI-supported technologies, these processes are sped up significantly, cutting handling time down to just minutes.

“Managing a shared Customer Service inbox can get pretty messy, with enquiries often falling through the cracks and frustrated customers on the other end,” said Aurélien Coq, Customer Service Product Manager at Esker. “ChatGPT analyses inbound customer emails, queries different systems and generates an answer for the CSR to use. This saves incredible amounts of time, freeing up CSRs to perform more fulfilling tasks, and customers get what they expect — a quick and helpful response.”

The responses proposed by ChatGPT are reviewed, edited if needed and approved by the CSR, who answers the customer(s) directly. Human supervision ensures customers still interact with other humans; ChatGPT simply provides suitable suggestions based on the available data to assist the CS teams with speedy and accurate responses.

The recent advancements in NLP based on AI technology now offer capabilities that comprehend and generate natural language text coupled with sentiment analysis. ChatGPT is valuable to CS departments for its ability to understand and respond to questions in a human-like conversational dialogue format. Due to the integration with other back-office systems such as ERPs, CRMs, transport management and warehousing systems, Esker’s solution understands emails through category classification actions, key data extraction and sentiment analysis. ChatGPT acts as a two-way translator between human natural language and system machine language and back between system language to natural language. It allows customers to communicate with internal systems in a controlled manner and supervised by humans.

“While not intended to replace a CSR, ChatGPT is an incredibly useful tool that improves the customer service workflows and the customer experience,” concluded Coq.

Bundling innovative AI technologies for over a decade now, Esker’s solutions are continuously evolving to address concrete business cases with Customer Service automation. By eliminating repetitive and error-prone tasks, automation enables CSRs to focus on building better customer experiences.

Esker will be expanding the use of ChatGPT for additional solutions within its Source-to-Pay and Order-to-Cash suites.

Read full press release here.

Esker UK

Unlocking Positive-Sum Growth with AI-Driven Business Solutions for P2P & O2C Cycles

Read more insights from Esker UK