How do you plan a project for a technology that’s constantly changing, with best practices that have yet to be established, for a critical business function that’s already struggling to keep up with customer demand?
Welcome to the discussion that’s going on (or that should be going on!) in every contact center. Everywhere.
You’re excited about the impact generative AI can have. Maybe because of the examples of extreme self-service with GPT that we shared during our Avaya ENGAGE session.
But you’re not ready to replace everything from the ground up. Obviously. There are big gaps in the tech, and best practices. But that’s OK, you can use the AI exoskeleton approach to begin building Gen AI into your existing platform.
And you’re still trying to get the ROI from the last platform migration. You’ve not fully leveraged the potential of classic conversational AI, let alone generative AI!
So what do you do? I’ve recently been helping a number of clients plan out the projects that will take them from old-school IVR to an AI-first contact center, and there are some repeating themes, so I put it all together into a high-level, cut-out-and-keep project plan that plots a course from old-school IVR to an AI-first contact center.
On the left is your contact center “Today” (labeled 1, in red, in the figure above). You’ve got some classic conversational AI (maybe the ever-popular Google Dialogflow or something like Kore.ai). It’s asking, “How can I help?” and doing some call routing. Some of those calls get handled by a legacy IVR app and/or go to an agent.
For your “next project” you’re looking to upgrade the call routing, improve personalization and authentication, and ultimately replace the legacy IVR apps with a more flexible, conversational experience. Just as you’d always planned once the migration was over. That’s steps 2 and 3 (again, labeled in red in the figure above).
But what about generative AI? You can’t ignore it, but it’s too immature, and evolving too quickly to plan confidently!
The answer, I think, is to get started, and be guided by data. You may not have an AI-first contact center yet, but you can build the processes, culture, and partner ecosystem now, and let the data – and experimentation – lead the way.
So as part of your next project, you should create an “AI innovation workspace” where your team, and others, including vendor partners, can collaborate to explore the capabilities of gen AI.
But don’t just ‘play around’. You need to be guided by data, so also make sure you’ve got a way to harness the “conversational data” that is being generated by customer interactions with your agents, IVR, and classic conversational AI apps.
During step 2, you’re rolling out a minimal viable product (MVP) of your classic conversational AI design: maybe the call routing, personalization, and authentication… these are usually the big hitters in terms of benefits and experience.
But while you’re doing that, you’re analyzing the conversational data, and experimenting with different gen AI tech in your AI innovation workspace. In the process, you might find a good use case for gen AI that is complementary to, or potentially replaces part of the classic conversational AI design.
So in step 3, you roll out a variant of the classic conversational AI design that leverages gen AI for some part of the automation. That’s easy to do, because you’ve got the routing, authentication, and personalization built-in already, so you’re literally just handing off some contact types to some new gen AI automation, instead of a human agent.
Now you’ve got a process up and running: Conversational Data + AI innovation à Gen AI use case à Gen AI production application. And then, of course, you get more conversational data, and you can continue with your AI innovation, and come up with more Gen AI use cases, that will ultimately generate more data. And the cycle continues.
Or rather, it should continue. But it will only continue if you switch from a project mindset to a program mindset. This should be your last AI project. From here on, it’s a program!
As I’ve discussed at length before… Gen AI is not like a better IVR. It’s more like better, cheaper, faster more scalable, and more consistent junior agents. You’re creating an AI workforce, that needs leading, and managing, and carefully monitoring and coaching and training.
That’s what I’ve depicted in steps 4, 5, and beyond in the above plan. Through the process of iterative, data-driven optimization, you’ll gradually replace all of your classic conversational AI and legacy IVR apps with gen AI. The investments Google, Microsoft, and Amazon are making into their generative AI ecosystems will empower you and your team with a gen AI suite, that makes it even easier to train, deploy, and optimize your gen AI experiences. Over time, you’ll automate pretty much everything that your junior agents do today, so you can escape agent churn, and begin running an AI-first contact center.
So that’s why this should be your last AI project. Going forward, you need an AI-first contact center program.
The AI-first contact center isn’t some dream of the future; it’s available now.
And if you missed our session at ENGAGE and want to learn exactly what an AI-first contact center is, and why it’s not just possible, it’s inevitable! You can watch a re-run of the event here.
