Companies are sitting on a wealth of valuable data that could be used to share relevant information with employees to improve the customer experience.
However, this data can often be buried in multiple systems, requiring staff to wade through reams of irrelevant information to find the precise detail they need, or work out the exact search term needed to access the crucial data. This process is time-consuming, meaning customer queries take longer than necessary to resolve.
Recent advances in generative artificial intelligence (GenAI) technology offer an opportunity for firms to offer a better customer experience by applying AI to knowledge management (KM).
Knowledge management software is defined by analyst Forrester Research as software used to create, publish and maintain curated content, letting employees answer questions, and customers find answers via web or mobile self-service. In the case of customer experience, employees are customer service agents.
With a solid KM foundation in place, firms can expect multiple benefits, according to Kate Leggett, Forrester Research vice-president and principal analyst for CRM and customer services.
“KM helps decrease operational costs by empowering customers to get rapid answers to their questions, which values a customer’s time,” she says. “It helps agents resolve issues more quickly, in a personal way. Both these benefits drive CX.”
KM supports adherence to regulatory compliance policy, ensuring customers get the one right answer to their questions; and it helps to drive customer engagement, ultimately helping increase customer retention and affecting revenue, says Leggett. By adding a GenAI layer to KM, there is potential for further benefits.
“AI is revolutionising knowledge creation and maintenance by automating and enhancing knowledge practices. Knowledge creation, improvement and sharing can be incorporated into core CX processes,” says Leggett. “GenAI-powered KM also makes it easier to access comprehensive information from multiple sources in real time. And knowledge can be co-created with GenAI, which speeds up its creation and helps generate new customer insights.
Adobe Population Health uses AI-powered KM to improve CX
Adobe Population Health is already using AI to support knowledge management and improve the customer experience. The healthcare provider serves more than 400,000 people via virtual and in-home services. The organisation offers a combination of medical care and social welfare support to address both health and wellbeing.
Many of its members are from underserved communities, and Adobe Population Health helps them access food, housing assistance and home modifications, such as wheelchair ramps, as well as providing medication and healthcare guidance.
The company is now using Salesforce Agentforce to support its clinicians and improve member care, mainly focusing on prompts. Prior to using the agentic AI platform, clinicians spent 15 to 20 minutes preparing for each member visit. This included confirming the latest information such as recent hospital stays or new medication, and verifying key medical details by checking past records, accessing multiple systems.
Using the Prompt Builder in Agentforce, clinicians can create templates that pull data from Health Cloud, and patient portals and databases via MuleSoft to generate care summaries in seconds rather than up to 20 minutes.
Adobe Population Health is now looking at the next stage for the technology, which will link knowledge management with the AI platform. The company has a broad range of educational materials stored in its systems and on its website, and would like to find a way to better share that with relevant members at the point of need.
Currently, when clinicians are working with a member, they will ask a series of questions relating to their health and social needs. If the member says they smoke three packets of cigarettes a day, for example, that could trigger a knowledge article or educational material to be sent to that member.
“Our clinicians and social workers are amazing – in their head, they have a lot of great information and resources, but a human can only hold so much. What I’m thinking about is how I can elevate our technology because, right now, it’s very much either the clinician has to pick, ‘I want this to be sent to them’, or ‘I’m going to pull this up myself’,” says Alex Waddell, CIO at Adobe Population Health.
“What we’re working towards is, as our members are charting, we have an agent that could say, ‘You just said that this person smokes, here’s some information on smoking cessation, here’s how you should be communicating this, offer this education material to them’.”
Waddell wants to get to the point where the system automation moves on from, ‘You answered diabetes – mail this out’, to where it’s putting that information in the clinician’s hands in the moment of the visit.
“So that we can, for lack of a better term, attack the problem in that moment. Because when [the patient] leaves, if you send them an education material, they might just throw it in the bin,” he adds.
While this is how the company is approaching a new, AI-driven era of knowledge management, Waddell is also considering how it could affect business processes.
“If we put our standard operating procedures in there, what if you could have an agent that could say, ‘You’re going in for this type of visit, these are the things that you need to cover, and here’s some material on how to do it. Maybe you’re going in to do a cancer screening with some device that you have, here’s how to use the device.’ We give them training, but it would be great to be able to elevate that information to them,” he says.
Waddell recently met with some of the Salesforce team to talk about an agent that could help further support clinicians. They discussed transcription and summarisation from transcription, which would make a huge difference for the nurses and doctors working for Adobe Population Health. When they’re carrying out a consultation with a member, they prefer text boxes for information rather than electronic medical records (EMRs) full of pick lists and dropdowns.
“It gets in the way of the focus with the members,” Waddell says. “We had talked about potentially building an agent that could listen and send data to the right places. So, if I say a member’s blood pressure is 129 over 89, I’ve got a place in Salesforce to put that data and that triggers off billing and different messages or knowledge articles or education material to get sent out.
“It’s just getting the system and the charting out of the way so that doctors and patients can focus on one another and solve problems,” he adds.
Lloyds Banking Group uses NICE to improve CX
Lloyds Banking Group is undergoing a major digital transformation project to strengthen its products and services for customers. As part of this, the company is using NICE’s GenAI technologyv, internally branded as Athena, to improve customer service by helping staff answer customer questions quickly and effectively.
Within 12 months, Athena has been scaled to support 12,000 customer-facing colleagues, who handle more than 25 million customer calls annually. The tool assists staff across a range of customer interactions, including personal banking services, fraud and disputes, and bereavement support. The company is now extending the tool with the aim of making it available to 43,000 colleagues by 2026.
Athena offers staff a user-friendly search tool that simplifies complex queries. It uses generative AI to summarise information and procedures based on detailed articles, which would otherwise take time for colleagues to read and digest.
“By speeding up information retrieval and comprehension, Athena cuts the time spent on these tasks in half,” says Suzanne Ellison, head of product – consumer relationships at Lloyds Banking Group. “This enables our colleagues to concentrate on what truly matters – meeting our customers’ needs and providing excellent service. By using Athena, we are combining human expertise with the efficiency of GenAI.”
According to Gina Whitty, director of product management at GoTo Connect, applying AI to KM is enabling its customers to take advantage of smart assistants that provide human agents with the up-to-date information they need to quickly, accurately and effectively meet customer needs. It does this while advanced receptionists are streamlining front-line services by automatically fielding simple requests using insights from knowledge catalogues and articles, whether that’s healthcare FAQs or queries about which parts manufacturers have in stock.
Further CX improvements are on offer through advances in AI-supported scripts, which previously were always extremely exact, using set keywords and intents.
“They established a specific formula that text-based chats and live calls had to follow, with human performance measured against how closely agents stuck to defined phrasing and protocols,” Whitty says.
But the free-flowing, fast-learning nature of GenAI is making it possible to move away from that rigidity. “Companies no longer have to invest significant time and resources in mapping out the ‘right’ approach for every potential scenario, which is driving a dramatic decrease in time to value,” she adds. “Projects that would once have taken at least half a year to complete, if not more, can now be kicked off, fine-tuned and rolled out in the span of three months or less.”
These benefits have led to a rapid spike in experimentation with GenAI that Whitty says is only likely to keep growing as more businesses begin to trust and appreciate its flexible abilities.
While GenAI provides businesses with CX improvements through its agile adaptation, its fundamentally data-driven nature also offers the benefit of consistency. The information companies enter into knowledge catalogues and articles create the parameters for smart agents to operate in.
“This means the services smart assistants deliver will be consistent because they are drawing from the same core database, which also allows for greater scalability,” says Whitty. “On top of that, it’s easier to ensure advice is always accurate and up-to-date by simply adjusting those data inputs.”
These efficiency gains can create significant cost savings in front-line resources, letting firms increase capacity without big training expenses. “And with the assurance that error rates will be low,” Whitty adds.
Virgin Atlantic uses Adobe’s CDP to centralise data
Virgin Atlantic can see the potential of AI-driven knowledge management, but is not at a stage yet where the business can take advantage.
The airline is starting to use Adobe Real-Time CDP to collate knowledge materials together, and sees a big opportunity to harness that combined with data from social platforms, which constantly generate vast amounts of content. This could lead to greater understanding of customer sentiment.
While it is too early for the firm to be testing out the application of GenAI to knowledge systems, Simon Langthorne, head of CRM at Virgin Atlantic, can see a point at which the business develops on-the-fly knowledge material, which then directly leads into a personalisation offering or a content change across the website, for example.
“From my perspective, the future is where we are, on the fly, creating personalisation use cases and you are able to manage those experiences, and those experiences have been created for the benefit of the customer,” he says.
The barrier to this at present is building trust in GenAI’s ability to create accurate materials. “It could be creating anything. So, how do I understand what it is creating so that I can trust it enough?” Langthorne says. “It’s knowing where the guardrails are when it comes to this, how do we create the guardrails around it?”
Leggett notes that with the rise of digital platforms, inaccurate or deliberately false information can spread rapidly, undermining trust in reliable sources: “The success of GenAI heavily depends on the knowledge quality used in training and deployment.”
Other barriers to uptake of GenAI for KM include findability and closed, siloed systems. “Knowledge is often scattered across different platforms, databases or departments, making it hard to locate. AI systems must be trained to quickly retrieve relevant knowledge to reduce staff time spent searching,” Leggett says.
“Meanwhile, teams often restrict information flow through access permissions and private stores of knowledge. Openness supports critical thinking, allowing agents to critically evaluate AI outputs and strengthening decision-making and overall business performance.”
For firms willing to tackle these hurdles, there is huge potential for CX improvements through applying GenAI to knowledge management, as evidenced by firms such as Adobe Population Health and Lloyds Banking Group.