Inside Genesys’ AI Studio: How agentic AI enables business. users to build better bots

Keith Shaw: Hi everybody, welcome to DEMO, the show where companies come in and show us their latest products and platforms. Today, I’m joined by Josh Goldlust. He is the Vice President of Product Management at Genesys. Welcome to the show, Josh. Josh Goldlust: Hi, thanks for having me.

Keith: We had a guest from Genesys on our Today in Tech show, and I said, “Hey, we’ve got to get you on DEMO and show some of the cool features.” So tell me about Genesys and what you’re going to be showing on the demo today.

Josh: Sounds great, and I’m really happy to be here. Genesys, while not necessarily a household name, is quite large in the global platform space. We're a global leader in AI-powered experience orchestration. That really helps companies—of all sizes—engage with their customers in more meaningful, personalized, and empathetic ways.

Keith: And the product you’re going to show today is called...?

Josh: We're going to show our newly released AI Studio and AI Guide. Everyone’s talking about agentic AI, so we’re really excited to demonstrate that.

Keith: Yeah, it feels like it's table stakes now. You come on this show—you’ve got to show something around AI. So who’s this designed for within the enterprise? Is there a specific role or department? Or is it meant for everyone?

Josh: We typically see two main user types benefiting from this. First are the folks creating virtual agents or bots. We've revolutionized how they do that—cutting down on time and eliminating the need for constant back-and-forth across departments. We’ve made it so that even non-technical business users can easily participate.

The second group benefiting is the end users engaging with these bots. Their experience becomes much more seamless.

Keith: And what problems are you solving for them? Is it about giving them an option they didn’t have before, or is it something else?

Josh: It’s incredibly manual today—how do you create a bot, understand the variables, and design the process? It’s often technical and slow. We're allowing people to use tech without being technical. Through a low-code interface, it’s now super easy for less technical users to get real results quickly. Keith: Okay.

All right, let’s jump into the demo then and see what you’ve got. Josh: Sounds good.

What I’m going to show first is the technical side—specifically for that business user. Most of us have interacted with virtual agents before—and let’s be honest, the experience is often poor. It’s very rigid: yes/no, decision trees, limited flow.

What we’ve created now is a no-code interface where someone can describe—in plain language—what they’re trying to accomplish. We're going to create a “guide” for a fictitious coffee company offering subscriptions.

So let’s say I want to set up a coffee subscription.

I know that when the virtual agent interacts with someone, it needs to collect three key pieces of information: Coffee type (what kind of beans) Delivery frequency Quantity (number of bags) So I’ll type that in plain English and hit “Create.” This will use large language models—specifically Sonnet—and take about a minute to generate.

Once done, it creates the full process, ready for tweaking.

It identifies all the variables and steps needed—frequency, quantity, coffee type—without requiring us to build the entire SOP manually. And because this is AI, summarization is built in. You can configure how conversations are summarized. For example, you might want bullet points or redact specific info. It’s simple and customizable.

Once the guide is built, we integrate it into a bot flow. That virtual agent can now reference this flow. So let’s say it first validates an address. All the needed questions are asked intelligently—not just dumped on the user.

Once validated, it seamlessly transitions into the subscription flow. What’s important is that this system empowers autonomous bots to get from A to Z quickly, but also gives businesses control.

Keith: Yeah, I’ve heard from others—they’re nervous about losing control or leaking data. Josh: Absolutely.

A great example is the Air Canada story, where the bot created its own policy and the company was held liable. We call that fully autonomous. What we’ve done is allow companies to balance traditional scripting with agentic AI. So some responses are scripted, others are dynamically generated.

Best of both worlds.

Let’s take a look from the customer’s side now. Imagine I’m on a website looking to start a coffee subscription. I engage with the virtual agent—it’s a great interface. I ask questions about the beans. Then I indicate I want to start a subscription.

Now it runs through the guide, starting with address validation. I enter “123 Main Street”—done. No need for multiple rigid steps. It intelligently collects all necessary info—delivery frequency, quantity, bean type.

In a traditional chatbot setup, building this experience would take a long time—and be rigid. Answer something out of order, and it breaks.

Or worse, it bumps you to a human because it can’t handle nuance. Here, I say “weekly,” “three bags,” “medium roast.” It handles it easily.

Now I test it by asking something off-script—“How does this compare to Starbucks?” The system doesn’t hallucinate. It respects the guide’s boundaries.

Keith: A lot of bots hallucinate trying to please users. It’s good that yours can say “I don’t know” when appropriate. Josh: Exactly.

Our research shows that 4 out of 5 customers want transparency and control. But only about 30% of enterprises currently have those safeguards. We’re excited to provide that.

Keith: Can I ask another question? Josh: Sure, but I’ll just go ahead and complete my subscription first. If I needed to, I could transfer to a human agent at any time.

Keith: Are most companies starting with simple projects, or are they diving into complex scenarios right away?

Josh: They want the complex stuff—that’s where the real value lies. Most traditional tools are very linear. Fine for basic tasks, but they fall apart if the customer says, “Oh, and one more thing…” That’s where LLMs and agentic AI really shine—handling branching logic, multitasking, and unpredictability.

The challenge is operationalizing that. With technical tools, it’s tough. With our platform—it’s much easier.

Keith: I know you’ve got more great features, but if companies are interested in this, where should they go? Do you offer a free trial? Josh: Absolutely.

Visit us at Genesys.com. There’s lots of info on AI Studio, Guides, and our broader platform.

Keith: Josh Goldlust from Genesys, thanks again for being on the show—and thanks for the demo. Josh: Thanks. This was great. Keith: That’s going to do it for this week’s episode. Be sure to like the video, subscribe to the channel, and share your thoughts below.

Join us every week for new episodes of DEMO. I’m Keith Shaw—thanks for watching.