We have seen in the previous article how companies can benefit from Agentic AI to accelerate sales. The key challenge in this would be the company’s ability to challenge its own sales strategy based on their customer behaviors in order to build adaptive tactics. This would require a deep understanding of its own processes and the ability to progressively connect its consumer data with operations. In this part, we will discuss the actual implementation of Agentic AI at scale.
Connecting the dots to unleash the full potential of Agentic AI
Agentic AI is basically the ability to power workflows so they become more flexible. It allows for more space for interpretation and automated adjustments, using GenAI capabilities to handle the arbitrations. You can already see where this will lead, as this means you are leaving the decision making to the GenAI engine—along with all the blind spots it might introduce, whether technical, ethical, or data-related. It’s not to say workflows are not subjects to any, but it means that there is a learning curve before you can be fully confident in this system. Once this is said and accepted, the starting point is to build a simple use case for Agentic AI.
Let’s take the Customer support business case as it’s the one that comes first in mind, with a lot of content that would be generated by customers reaching out because of problems that would be described with natural language. In this setup, the Agentic AI system needs to be able to filter, compartmentalize and ultimately define what type of issue needs to be addressed. Then it would tap into the pool of existing knowledge to have a conversation with the actual client to move forward. Many options are on the table at this stage:
Providing links to the FAQ: the easiest way to answer, but with the least added value for the customer and definitely no need for any Agentic AI for such a project
Reviewing the issue with the customer to make sure it is well understood and there are no blind spots, making it possible to move forward with either a human agent or generating a ticket for a technical handling
Explaining, autonomously, how the issue will be addressed by the company and what can be expected from a customer standpoint depending on the customer profile and background
Obviously, the 3 actions should be done in parallel as it reinforces the acknowledgment of the issue for the customer. Still, process wise, those are 3 types of operations that are vastly different in terms of complexity if you want them to be relevant for the customer. Keep in mind KPIs such as your CSAT and NPS are the ones that will be under pressure if you’re doing poorly here. And the more ambitious you want your answer to be, the more data you would need to connect, meaning the more agents would need to be involved in your Agentic AI use case.
Agentic AI for Customer Care: leveraging CDP, NBO/ NBA, tech support, sales, CRM and omnichannel at once
The dots that need to be connected with the Customer Care/ Support business case are quite obvious yet very complex. To provide a relevant answer and strengthening brand affinity with your customer, you need to understand where she comes from, her lifetime value, the current stage she is within your purchasing funnel and loyalty program. You also need to anticipate the potential outcome of the issue she raised to manage expectations, meaning you need to connect with the operations and also the CRM and commercial part for resolution.
Mapping out the dots to connect is the easy part, as now you need to actually connect the databases and build the “window of freedom” or boundaries within which the agents need to work and collaborate with each other. Easier said than done as explaining the human workflows to AI agents isn’t that simple as of now. This is why we are still at the beginning of the story with Agentic AI. You now have a rough understanding of the scope of work and can have a feel for the investments that are required. It starts with the data quality to make sure you’re building on a reliable and stable ground.
This is why the recommendation is to start small, with a few operations to deal with, in order to be able to adjust and correct whenever needed. And scale from there. Otherwise you’re in for a challenging project where you’ll be parallelizing a lot of initiatives that need to come together at some point without any intermediary milestones. But the beauty of Agentic AI is that once you are comfortable with your small scale project, you can expand faster as you are building on GenAI assets, and potentially ML assets, to increase the pace of growth of your operations.
Agentic AI to build a stronger organization and improve your operational excellence
Building such an ecosystem able to leverage such a large amount of data is always really challenging. As you may have anticipated, this topic is at the crossroads between the data management capabilities and agility and the data driven culture of your company. As you will be introducing a new “individual” in your task force that needs to be trained before being able to perform at the required level, the current task force will need to pivot in order to adjust to the newcomer and not the other way around. And this will put a lot of pressure on the whole organization.
This Agentic AI wave is definitely challenging the current organization processes at both the tech and organizational levels to deliver results. Digital born companies have this in their DNA and are moving fast because of it thanks to a culture that is already data driven from day 1. This is the gap that needs to be addressed, the technology gap will be managed eventually since that is only a matter of connecting the existing data with each other. You will encounter problems down the road as it will also highlight some business process issues that were dormant and never addressed but that will be adjusted somehow.
The real question to ask yourself is how to move step by step, identifying the opportunities for quick wins while building the tech platform you need to run your operations and transform the company’s culture one project at a time. Because Agentic AI is, before anything else, about your organization and how it functions within a data-driven environment, it will inevitably reshape your operations and decision-making processes. It will have a massive impact on the way you operate and how you create value as it will provide you with a clear and critical analysis of your strategy.
Next and last chapter will be about KPIs to track when building your Agentic AI operations. See you next week.
Leave a Reply