In a move that underscores its commitment to AI-powered customer experience innovation, Genesys announced Genesys Cloud AI Studio on June 25, 2025. This new AI command center is described by the company as a foundational step toward enabling organizations to implement agentic AI-driven customer engagement responsibly. In conjunction with Genesys Cloud AI Studio, Genesys announced the first major solution for this capability, Genesys Cloud AI Guides.
Before we dig into the new offerings’ specifics, let’s look at how Genesys Cloud AI Studio fits into Genesys’ broader AI strategy.
Tony Bates, chairman and CEO, Genesys, shared his vision for a future where customer and employee relationships are transformed through a human-centric approach to developing and implementing AI at Xperience 2024, the company’s annual customer event. He presented a framework describing “levels of experience orchestration,” which organizations typically go through as they implement self-service and automation in their contact centers. Bates and Peter Graf, senior vice president, strategy, Genesys, subsequently published a blog that described the six levels of orchestration.
-
Level 0: Zero orchestration (foundational, manual interactions).
-
Level 1: Basic menu-driven interactions using IVR technology.
-
Level 2: Predefined dialog automation across channels.
-
Level 3: System-generated conversations using large language models and generative AI.
-
Level 4: Empathetic experience generation with emotionally aware AI.
-
Level 5: Universal orchestration with fully personalized, seamlessly integrated experiences.
Genesys’s view of the levels of experience orchestration continues to evolve. During an analyst pre-briefing on the AI Studio announcement, Olivier Jouve, chief product officer, Genesys, presented the graphic below, slightly redefining levels 4 and 5 and incorporating the term “agentic.”
Genesys
As discussed in my post, “AI Agents Are All the Buzz at CCW and Cisco Live,” earlier this month, recent announcements from CX vendors focused heavily on delivering new AI agent solutions. It is not surprising, then, to see Genesys evolve its point of view on levels of experience orchestration to align with the rapid market response and move towards agentic AI – an agentic zeitgeist.
Genesys Cloud AI Studio is a platform designed to be a centralized hub for building, managing, and scaling AI faster across an enterprise. It provides organizations with tools—embedded with guardrails, permissions, privacy, and other trust elements—that allow for the creation of intelligent virtual agents (IVAs) and additional Genesys AI solutions that are smarter and deliver more personalized and autonomous experiences than previously possible.
In an industry analyst pre-briefing, Rahul Garg, Vice President of Product, AI & Self Service at Genesys, described Genesys Cloud AI Studio as a new workbench where Genesys Cloud users can configure the AI tools available across the Genesys Cloud platform. The activities that Genesys Cloud AI Studio has the potential to support include, but are not limited to:
-
AI Guides: train AI Agents with natural language instructions (discussed below).
-
RAG (retrieval-augmented generation): RAG is an approach that combines the strengths of information retrieval systems with generative language models. Instead of relying solely on a pre-trained language model’s internal knowledge, e.g., LLaMA or AWS NOVA, AI Studio can assign Genesys AI applications (e.g., to use relevant information from sources like databases or documents) to enhance the quality and accuracy of generated responses.
-
Sentiment & Empathy: LLM-derived customer and agent sentiment.
-
Intent & Topic Mining: extract intent and topics from conversations.
-
AI Scoring: automate evaluation effectiveness.
-
Anomaly Detection: auto-detected behaviors, triggers, and alerts.
-
Summary Model: customize auto-summary outputs.
When an analyst asked when the full range of AI Studio’s capabilities would be available, Jouve responded that “most of this is available today and, with continuous delivery, is updated on an ongoing basis.”
“The goal is to bring it all into one place,” Garg added, highlighting how AI Studio will centralize AI capabilities previously scattered across the platform. Garg showed analysts a demo of two capabilities: AI Guides and Custom Conversation Summaries.
AI Guides, the first capability released for Genesys Cloud AI Studio, allows users to train AI Agents with natural language instructions. A no-code interface enables organizations to build and deploy IVAs that can reason and act within configurable guardrails defined by the business. The goal is to allow brands to rapidly design, deploy, and govern virtual agents to operate with greater autonomy across complex, enterprise-wide customer journeys.
The features of AI Guides demonstrated included:
-
Natural Language, No Code Required.
-
Build Once, Deploy Anywhere: Experiences designed once can be deployed across IVAs, Genesys Agent Copilot, and more to maintain consistency and reduce duplication of effort.
-
Enterprise-Grade Collaboration: As is true of other tools being released in the customer experience market in 2025 (e.g., Talkdesk, Five9, and NiCE), one of the goals of AI Guides is to allow connection of front, middle, and back-office systems to execute tasks and automate workflows.
-
Built-in Guardrails: Configurable, testable safety controls to help increase accuracy, appropriate tone, and policy compliance to support the responsible adoption of agentic AI.
A key aspect of Genesys’ approach to AI is its focus on agentic capabilities. During the analyst briefing, I asked Jouve: “What is Genesys’ definition of agentic AI?”
He said, “Agentic is when we start moving from deterministic to non-deterministic. If something is scripted, it is not agentic AI. I think the ability for your AI to start doing some reasoning and making some decisions in an either semi-autonomous way or fully autonomous way, is when agentic is taking place.”
Interactions through level 3 in the Genesys orchestration model above would be considered deterministic – developers have created every possible question and response. In levels 4 and 5, customers can ask, and the AI agent can respond, with ad hoc language.
Jouve notes that while agentic AI involves non-deterministic and non-scripted capabilities, it still must respect guardrails.
This definition of agentic aligns with Genesys’ adjusted levels of orchestration, where level 3 solely uses generative AI in self-service agents and level 4 and 5 move to a more agentic experience.
Autonomous actions require the use of reasoning models. Earlier this month, a flurry of discussion about AI reasoning capabilities was occasioned by the release of a research paper by Apple, The Illusion of Thinking: Understanding the Strengths and Limitations of Reasoning Models via the Lens of Problem Complexity. One of the paper’s conclusions was, “Our findings reveal fundamental limitations in current models: despite sophisticated self-reflection mechanisms, these models fail to develop generalizable reasoning capabilities beyond certain complexity thresholds.”
The Wall Street Journal, among other business news outlets, implied that Apple’s findings run counter to exaggerated claims about AI capabilities. They noted that models failed logic puzzles solvable by children despite extensive training.
When asked how Genesys addresses reasoning in Genesys Cloud AI Studio, Jouve acknowledged the limitations of current large language models while highlighting their progress. “The report from Apple is spot on about the full capability of reasoning…. The large language models are not there yet. They are doing simple reasoning – that goes beyond something that would be scripted, like a decision tree. We are beyond that. But we are not delivering full reasoning yet.”
Jouve drew an analogy to autonomous vehicles: “Do you fully trust a car to be fully autonomous? We are getting there,” said Jouve. “It is the same for the contact center. Do you believe a self-service agent is completely autonomous and makes some complicated decisions? Think of a call center agent in an airline, being able to do more than rebooking, but finding uncharted solutions to get you out of the airport… we are getting there. But the large language models do not support full reasoning yet, so it would be wrong to say that.”
I appreciated Jouve’s transparency and insights in answering my questions, and I am sure Genesys customers do, too.
Availability of Genesys Cloud AI Studio capabilities and AI Guides are expected in the second quarter of the company’s fiscal year 2026 (May 1-July 31, 2025).
