A “How To” Guide for Customer Journeys and Agentic AI

The year has been building with great momentum. I spent the last week in Amsterdam at the inaugural Institute 4 Journey Management, engaging with other experts in the Journey Space. It was an opportunity to see what the leading companies are busy with and the successes that they are achieving. Feedback from key industry analysts is that they have never seen more interest in Customer Journey Management. However, many companies are still grappling with how to embrace Customer Journey Management.

What has equally been a significant point of discussion in the business social media space (and specifically LinkedIn) is extensive discussion about agentic AI and more recently MCP – Model Context Protocol.

I am going to quick start off to position AI Agents. This is going to be more of a layman’s explanation and thus my technical explanation might not meet muster.

An AI agent is a combination of an LLM model (eg ChatGPT), coupled with a set of instructions and a set of tools that can do something. For example, you might have an agent that can:

• Search the web for offers on car insurance
• Another that can compile all the searched information into comparable options
• Another that has a set of tools that can potentially populate quoting forms and retrieve the quoted prices
• Another than can provide guidance on the pros and cons of various car insurance options

Each of these agents (consisting of an LLM, set of instructions, tools for interacting with the environment (for example quotes) can then be controlled by another agent to get a job done.

One of the recent developments in this space is a standard called “Model Context Protocol”. What is really important about the MCP protocol is that it is developed and published by the service provider (for example an insurance company that provides quotes) so that the agent can interact with external services in a standard way. Think of it as an Appstore for tools that agents can use.

Other example of MCP tools could include:

• The ability to show availability at a hotel
• The ability to specify how a claim should be submitted and the to collect that information
• The ability to show the best mobile phone contracts between certain price ranges

MCP allows AI Agents to access real services so that the agents can complete their tasks. MCP is being endorsed by many sectors of the AI community and the large players such as OpenAI and Anthropic.

When developing a AI solution (made up of several agents as above), it is possible to hand code these agents but there are really excellent frameworks such as Langchain and Semantic Kernal that speed up the development of agents and then there are visual tools that use the AI frameworks to allow non-technical people to develop agent solutions. The development in this space is happening at lightening speed!

Given our focus on journeys, the key point of interest is how to plug in an agent solution into a journey framework and what a journey framework offers to enhance the capabilities of agents. I would like to move onto that next.

When we deliver a journey orchestration and journey analytics solution, all journeys have:

• A start
• A goal
• Tracking of progress towards the goal
• In the event of a failure, drop offs
• The ability to collect insight about the customer along the journey

A wonderful opportunity arises when the world of agents and journeys come together for these reasons:

• Agents work best when there is a goal. They are relatively intelligent but given that general AI has not been achieved (and likely won’t be for some time), having a clear goal optimises the use of agents
• Agents can be dramatically enhanced when they have better context. In the example above, the more the agent knows about the needs of the customer shopping for insurance, the better the fit between what on offer and what the customer needs
• The extraction of data from corporate systems can be challenging but journey analytics and journey orchestration solutions are optimised to do that, and they have well developed data extraction and mapping capabilities to be enable the visualisation and analysis of journeys
• Agent solutions are typically absent an effectiveness and measurement framework and customer journeys provide exactly that.

The below image provides a view of how this is done:

Customer Journeys bring a unique perspective to Agents in that the concept of time is introduced. The agent is no longer just waiting for the customer, it an guide and nurture the customer along the journey because there is a framework for assessing progress in place. It can thus nudge the customer when they get stuck, making sure the agent both meets the needs of the customer but also keep up the momentum of the journey (with many customer journeys spanning days)

The future of customer journeys and customer experiences will be infused with AI agents. All the technology to do the above is in place and we are going to see a massive groundswell over the next few months. I hope that the above newsletter provides a bit more insight into the “how” where a lot of I have been reading is far broader and absent the detail.

I look forward to feedback and debate on the topic.

Have a brilliant week ahead and as always enjoy the journey

Trent