With the buzz around generative AI growing louder by the day, many CX leaders are already looking to incorporate it into their automated customer support strategies. But knowing what you need to do is not the same as knowing how to do it. That’s why we recently hosted a webinar to talk about the practical steps businesses should take before they adopt generative AI for customer support automation to ensure they can make the most of it.

To lead this discussion, Ada’s CEO Mike Murchison is joined by guest speaker and lead machine learning (ML) analyst at Forrester, Rowan Curran, as well as our CPO, Mike Gozzo. Let’s dive into the key points they covered so you can get an idea of what you’ll learn — and why you definitely want to watch the full session.

The state of generative AI

Rowan kicks off the webinar by providing an in-depth look at the current state of generative AI, including recent developments and where we might be headed in the next 5 to 10 years. From rapidly evolving large language models (LLMs) to text-to-image generative models, he shares insights into what generative AI can do right now and what type of inputs are needed to generate the kind of content consumers are looking for.

From there, he turns to how we can put LLMs to work in the enterprise.

“Base models understand the world, but they need to understand your world and do it succinctly.” — Rowan Curran, Lead AI/ML Analyst at Forrester

Rowan walks us through various LLM model types and use cases as well sharing actual examples of AI-generated content to illustrate its strengths and weaknesses. He also gives examples of various types of natural language prompts that can be used to train the AI model.

Getting ready to adopt generative AI

Next, Mike and Mike take the mic (see what we did there?) to talk about the practical steps organizations can take right now to prepare — and why it’s vital that you embrace this game-changing technology.

“It’s absolutely critical to recognize that AI has just had its Netflix moment. For CX leaders, getting left behind puts you at risk of becoming Blockbuster.” — Mike Murchison, CEO at Ada

Mike Murchison says the best way to approach generative AI is with a human perspective: onboard the model just like you would any other new hire. The more thoroughly you prepare the AI, the better it will perform from the get go. There are 3 things you need to teach it how to do.

#1: Teach it to read all your company documents

This is a great opportunity to make sure you have all of your documentation in order. Start by auditing documented knowledge across the business to ensure it’s truthful, correct, and up to date — not to mention complete. Documentation can include knowledge base articles, product manuals, internal guides, and historial transcripts from top agents. Anything that will help the AI learn about your business and your customer needs.

“Teaching generative AI to read is akin to making sure your CX agents are trained with the detail, nuance, and breadth of information they need to serve your customers properly.” — Mike Gozzo, CPO at Ada

Once you’ve surfaced and organized everything the AI can use to generate content, you also need to validate it for accuracy. The more accurate your documentation, the less likely the AI will “get creative” with the facts — which can be damaging to your brand.

#2: Teach it to perform actions

According to Ada’s internal customer data, roughly half of all resolutions require an agent or chatbot to do more than simply answer questions. So if you want to make the most of generative AI, you need to teach it how to do things.

Just as an employee must be able to bring together information systems from across your business to do things like prepare returns, reset passwords, and more, the AI also needs to access data and trigger actions across various systems of record. And that requires APIs.

“If you don’t already have an API strategy, you need one now.”
— Mike Gozzo, CPO at Ada

Start by auditing your tech stack and making sure all the critical systems your contact center agents are using have an API available. You need to understand what the API can do, how well it’s built, and what kind of support is available from the vendor. If most of your software is cloud-native, you’re probably okay. But if you’re still using legacy or on-prem solutions, you may need to invest in some additional layers here to enable automation.

#3: Teach it to get expertise

In addition to connecting with various systems, you’ll also need to connect the AI with experts across your organization to help it solve the most challenging customer issues. Again, the better you prepare the AI, the better it will perform.

“Humans are an important part of AI strategy. Expertise exists across the organization and you need to identify it and integrate it into this initiative.”
— Mike Gozzo, CPO at Ada

Start by taking stock; map out your organization to understand what nuggets of knowledge exist, who holds them, and where. Don’t limit your search to agents or even the customer support organization. For example, members of your finance team might have critical knowledge about certain customer needs that can be used to coach the AI.

Make sure every subject matter expert is equipped to review and augment automated interactions. This is an all-hands-on-deck moment, so include any individual or team that can provide insight or expertise in the initial deployment and ongoing operation of generative AI.

Get your checklist

To dig deeper into each of these areas, learn practical tips, and see a demo of what generative AI can do, watch the full webinar here . Bonus: as soon as you sign up, you’ll get access to a downloadable worksheet with actionable steps that you can take to start preparing your CX organization for generative AI. Don’t get left behind — get moving.