How, Practically, do AI Agents, fit into Customer Experience Management?

The press is filled with articles on the role of Agentic AI in impacting customer experiences. The hyperbole is along the lines of “agents will act autonomously and solve customer problems”. There are two perspectives on this:

• Yes, sounds fantastic
• How on earth does that happen in the real world?

We have been working on the above solution, with a strong focus on the second point. I would like to use this newsletter to break down the problem and in a bit more detail and discuss connecting the dots between the grand vision and the practical reality.

AI Agents are a great concept – to have some higher level intelligence independently act on your behalf to get whatever you need to get done, done! The first practical aspect we must deal with is to be specific on the “what” that needs doing. In the inQuba world, we specify that as the goal of the journey. The goal defines that success has been achieved. It is a point on a journey, and it can have certain measures like “a great experience” associated to the goal point. Importantly, different people have different ways of getting to the goal and thus the path of a customer towards the goal point can be very varied and unique to individuals.

To progress along a journey to the goal there are a few important constituent components:

• Progress must be occurring (progressing through the journey steps) and getting the work done to get to the goal
• Making sure the right information is provided to the customer so that they can assist in whatever comes next
• Assessing the perceived value delivery and educating and engaging the customer
• Assessing the ease or difficulty and assisting if blockages occur

AI Agents can help with each of the above components:

1. Driving Progress

Progress along a customer journey is often centred around information, particularly for financial service products and every for services such as telecommunications and hospitality. The often means information must be shared or information has to be collected. AI Agents can assist on both fronts. Agents can adjust engagement as a function of progress. Nudging progressively harder where the customer is lethargic (and not dissatisfied) but supporting where the customer is getting stuck and clearly needs help (potentially dissatisfied or confused). With respect to information collection, when coupled with chat channels such as email and WhatsApp, information that includes multimedia can be ingested and then translated or transcribed into the correct format for the backend systems to process (for example from voice to text).

2. Providing the Right Information at the Right Time

Given the low attention spans of people in our world of multiple channels, a plethora of media and our busy lives, customers only want the information relevant to their situation. As they progress the information they require will change. By having AI Agents extract the relevant information for the situation, it reduces the burden of the customer going through the journey. The attention span and effort required is reduced, making the journey feel easier and less cognitively demanding. This naturally requires a very rich view of the context of the customer both in terms of more permanent attributes (such as customer segment) as well as short term experiences (such as current frustrations reflected in feedback).

3. Assessing Value Delivery and Educating as Necessary.

If a customer perceives value, they are more likely to make the effort to progress. The value perception needs to be assessed, and the education and assistance provided increased or decreased based on perceived value. If the customer is not perceiving value, the action might be to escalate to a human to assist (in the case of a large value gap) or if the gap is small, then personalised content might do the job. There may be situations where the choice might be to incentivise the customer with an appropriate reward to facilitate progress. The power of small incentives to drive behaviour have proven to be a very powerful tool.

4. Removing and Managing Blockages and Frustrations

Blockages can be observed by idles (lack of journey progress) where there is engagement (differentiating this situation from lethargy). If a customer is actively engaged (indicated by interactions with the channels) but not making progress, then they are likely frustrated and need assistance. In instances where customers are clearly dissatisfied with their current situation, it is appropriate to assist with that frustration often by having key people assist. Idles are the obstacles that nudges aim to overcome.

Bringing it All Together

In each of the four scenarios above, there is a customer context. This can include aspects such as:

• Progress along the journey or lack of progress along the journey (idles)
• Customer feedback and sense of each and frustration as well as value perception
• Where exactly the customer is on the journey (started or near the end)
• The customer’s individual situation (for example the customer is not tech savvy)

There is also the requirement to have content prepared and organised that is relevant to that part of the journey. For example, the information required at quoting stage is very different to the information required at application stage. The information arrangement is key to make sure it is well organised ahead of customer engagement.

The key ingredient to make sure that there is definitive progress along the journey is to have a level of intelligence that can deal with all the above information, make decisions and drive action. There are a lot of moving parts to get to make sure that the customer is most effective nudged along the customer journey. A well-trained AI capability provides the capability to:

• Determine when to nudge more and when to slow nudging
• What information to provide when to the customer
• Process multimedia along the journey (receiving or sending)
• Recognise when there is perceived value delivery and when there is not
• Recognise frustrations and blockages and how to overcome those frustrations and blockages

Customer journeys have many moving parts and only by having a model that can arrange data in a coherent view of the customer journey, can the data be managed and the actions orchestrated.

AI Agents are an incredible opportunity to assist customers in reaching their goals. However, they need to be supported by a strong data model and an infrastructure that facilitates engagement along the journey.

In considering the implementation of your AI Agent strategy, here are several questions that need to be answered:

• Do you have a model for arranging customer data so that you can make sense of each customer’s journey?
• Can the data model feed into the AI Agent so that it can make decisions?
• Can different actions be taken depending on the context of the customer?
• Can you measure the effectiveness of your AI Agents based on their recommendations?
• Does the infrastructure allow for rapid prototyping and refinement of the AI Agents to optimise results?
• Can your agents start the customer on a new journey giving context to the objectives of that journey?