Zendesk’s New AI Resolution Platform Aims to Tackle Customer Service’s Biggest Question

Zendesk CEO Tom Eggemeier recently faced a common customer service frustration. While traveling, he used an airline chatbot to rebook a missed connecting flight before losing connectivity. But when he landed and reconnected, the virtual agent had no memory of the conversation. Worse, the human agent he eventually reached had no record of it either. Frustrated, Eggemeier called it a “lose-lose situation.”

He explained that everyone lost because not only did Eggemeier’s issue not get resolved to his satisfaction, but the airline company also lost money because of the agent interaction. This scenario represents the issue Zendesk is looking to tackle with its newest software offering: helping companies figure out if they’ve actually addressed their customers’ problems.

Unveiled at Zendesk’s Relate conference, the Resolution Platform is designed to ensure every customer issue has a clear path to resolution. This no-code suite includes AI-powered agents, a comprehensive knowledge graph, automation tools, enhanced quality assurance, and improved governance monitoring, all aimed at streamlining service operations.

“The only metric that matters in customer service is resolution,” Zendesk Chief Executive Tom Eggemeier said in a statement. “The Zendesk Resolution Platform is not just making service faster—it is making Agentic AI actually work for service, solving every issue with less effort and better outcomes.”

He continued, “Our network of AI agents built with service at the heart works like a well-trained search and rescue team, ensuring every interaction leads to a resolution. And as the only large service software provider offering outcome-based pricing. We make sure customers only pay for problems that are resolved—not for interactions or failed attempts.”

Disclosure: I attended Zendesk's Relate conference as a guest of the company with my flights and hotel covered. Zendesk did not dictate the contents of this post. These words are my own.

What Makes Up the Zendesk Resolution Platform?

Zendesk’s new Resolution Platform consists of AI-powered tools designed to automate issue resolution, streamline workflows, and improve insights. The suite includes AI agents, a detailed knowledge graph, automation and connectivity tools, governance controls, and advanced analytics.

Jon Aniano, the company’s senior vice president for product and CRM, described these components as being “critical for a complete resolution platform, and we want to make sure that AI works together with humans solving every issue with less effort and better results for your customers.”

AI Agents

Zendesk is introducing off-the-shelf intelligent bots that have been built to improve service delivery. Reetu Kainulainen, the vice president of product management who joined through Zendesk’s acquisition of his startup Ultimate, calls these agents “mission-driven and goal-oriented.” They’re powered by adaptive reasoning, meaning the bots can analyze every request, identify the best course of action, and adapt on the fly while adhering to the company’s processes. Organizations won’t need to worry about manual conversation design.

There’s also an agent builder to enable organizations to develop specialized bots tailored to their needs. Like similar platforms from Salesforce and Microsoft, this requires no coding ability, just natural language prompting.

Kainulainen commented that after ten years of building AI-based platforms, he knows making an agent is time-consuming. “You can build a really bad one or a really good one, depending on your skill level. And this has been a blocker for many companies, because they don’t have time for this. So the new AI engine builders we’re announcing makes it incredibly easy.”

Aniano was more enthusiastic, hailing it as a “paradigm shift” in which bot-building was moving away from the need for detailed workflows to simply “describing the customer service processes, visualizing the adaptive reasoning that the bot is going to take when solving problems, and letting it resolve customer issues when it connects to all the different systems.”

Lastly, Zendesk is upgrading its Copilots. Initially announced in April 2024, these intelligent guides use past customer service experiences to help with workflows, address customer needs, and make recommendations on future interactions. Eggemeier tells me the company’s Copilots usually see around a 20 percent uplift in efficiency: “They’re more accurate, they get better customer satisfaction.”

Now, they can be trained using Zendesk knowledge sources and third-party systems to self-execute business procedures on behalf of agents—without needing a line of code. Copilots also now support action flows, enabling them to automatically execute actions on non-Zendesk systems like Jira and Slack to generate responses. There’s also a feature allowing the admins to specify to Copilots when it’s time for a human agent to take over the conversation.

Knowledge Graph

As Data Cloud is at the heart of Salesforce’s Agentforce, so too is the Knowledge Graph for Zendesk. With it, companies can index and connect their knowledge sources, thereby creating a semantic layer of the data that’s used to train AI agents, Copilots, and other AI-powered solutions. Today, Zendesk said it powers more than 50,000 such help centers.

With the Resolution Platform, Zendesk is launching its Knowledge Builder tool. Using generative AI, admins can quickly generate a “ready-to-use” knowledge base. Kainulainen points out that the days of trying to write help articles and content based on guessing customer queries are over. Instead, Zendesk will leverage a company’s unique context—its brand, business operations, and existing knowledge inside and outside the platform—to generate a complete knowledge base with a single click instantly.

Finally, Zendesk is deploying a generative search feature that displays answers at the top of the help center, eliminating having to sort through links on the page.

Actions and Integrations

“It’s great to have AI agents helping your customers resolve problems automatically. It’s great to have AI agents helping your human agents deliver better outcomes for your customers, but in order to drive real resolution, you actually need to take action,” Aniano remarked.

To that end, Zendesk’s Resolution Platform features two builder tools enabling admins to create their own agentic workflow and develop custom applications with a Gen AI builder.

The first, Action Builder, helps teams tackle complex problems and is a no-code solution. Organizations can integrate AI and human agent workflows across any system using the tool. It also includes connections to some popular tools Zendesk’s customers connect to, such as Slack, Jira, and Salesforce. “We’re talking about pointing and clicking, creating your logic flows, exposing these actions to AI agents for automation and exposing these actions to human agents to Copilot, and connecting all the different systems it takes to resolve the customer problem, all with a low code no code builder,” Aniano explained. The goal is to allow organizations to do more within the Zendesk environment without spending on other vendors or custom development.

The second builder is also a no-code solution called App Builder. Using natural language prompts, customers can develop custom apps in Zendesk. Similar to GitHub Copilot, Bubble, and other AI-powered code development solutions, Zendesk’s App Builder lets organizations produce their own customer experience app built on the platform’s framework and then distribute it through Zendesk University.

Governance and Control

Zendesk claimed it wants to make it easy for teams of all sizes to maintain the privacy and security of AI tools. Like its other products, the Resolution Platform has tools that allow organizations to ensure data accuracy and effectiveness. The first is Advanced Data Privacy and Protection, which provides additional data safeguards for customers in sensitive industries or have specific corporate controls. This includes access logs, advanced data retention, data masking, sophisticated encryption, and advanced redaction.

Another tool is Automatic Redaction for Zendesk Voice. Organizations can deploy it to redact personal data from call recording transcriptions automatically. This is timely as Zendesk is redefining what a traditional call center is and has expanded its AI agents to voice channels.

Finally, Zendesk is introducing Reasoning Controls, a feature that enables admins to remain in control of AI agents. This feature monitors how these bots solve problems and allows them to refine their behavior to ensure resolution confidence.

“It’s not just about security, it’s also about transparency,” Aniano asserted. “What is the AI going to do? What chain of reasoning is it going to use? There are many AI tools and AI models out there that are just black boxes. There’s little insight into how they work. You don’t know what they’re doing behind the scenes, but our recent controls actually give customers full transparency, so you’re gonna be able to see what the AI agent is doing, what in their decision-making process. And I think that’s really important, because once you know what the AI agent is going to do, you can actually adjust that behavior.”

Measurement and Insights

Analytics make up the final component of the Zendesk Resolution Platform. It consists of AI-powered tools designed to help admins manage their AI operations, streamline reporting, and turn insights into actions. The Resolution Platform has four such services:

The first is the AI Insights Hub, an operations center admins can use to receive onboarding advice, key usage metrics, and recommendations. It’s the command center where organizations can oversee all their Zendesk-powered AI agents, tracking what’s working with the bots and learning how to make their Copilots more efficient.

The second is Recommendations. This feature gives AI-powered ideas on which jobs to be done should be automated. The third is Prompt-Based Reporting, giving admins an AI analyst that will generate reports, identify the appropriate data, and deliver insights using conversational prompts. Aniano claimed this will be a critical area for the company moving forward.

The last piece of the measurement and insights puzzle is Custom Quality Assurance (QA). Using natural language prompts, admins can tailor the AutoQA scoring and Spotlight discovery to meet their needs while identifying potential red flags in customer interactions.

“This is huge in the world of QA…It lets you tap into generative AI and comprehend insights. You can extract behaviors out of what’s going on inside your customer service tickets using generative AI,” Aniano explained. “One of the things our customers love to use this for is looking for churn risks. So for each interaction, you run this scorecard, and you can say, ‘Okay, it looks like this customer may churn,’ and that’s very expensive. So we can then go and find those insights, talk to that customer, reach out, and potentially turn that situation around. So having custom QA on top of all the interactions happening inside of your customer service center is a pretty critical part.”

How Zendesk Sees This Fitting In

“Zendesk’s Resolution Platform is really the culmination of all of the products that we’ve been building over the past several years,” Lisa Kant, the company’s senior vice president of product and solutions marketing, replied when asked how Zendesk’s newest product compares to those from competitors. “We believe the Resolution Platform is fundamentally different than what’s out there in the market today.”

She explained that the solution for customer resolution isn’t simply to apply AI. “The problem is that AI is just not enough,” Kant pointed out. “We hear time and again from prospects that they are struggling to see a return on investment…they’re struggling to get AI to actually work, and that’s because we believe AI has to be part of an integrated system that combines AI, human capabilities, actions and workflows, and most importantly, this integration of this insights and measurements layer that is ensuring that the system is learning and improving over time.”

Aniano echoed the sentiment, replying that “one of the things that AI has done to software is it brings the actual outcome of the software much closer to the customer and the buyer. And so we know when people buy Zendesk, what they really want is resolutions to their customers’ problems, whether that’s automated, whether it touches external systems, or just touches a knowledge base. They just want the customer’s problem to be solved. They want that resolution.”

When Will This All Be Available?

• AI Agents: Currently available through Zendesk’s early access program (EAP)
• AI Agent Builder: Currently available through the EAP
• Expanded Knowledge Sources: Generally available (GA)
• Action Flows: EAP
• Instructions: GA
• Knowledge Builder: Available through the EAP starting May 2025
• Generative Search: GA
• Action Builder: Available through the EAP starting April 2025
• App Builder: EAP
• Advanced Data Privacy and Protection: GA
• Automatic Reduction for Zendesk Voice: GA
• Reasoning Controls: EAP
• AI Insights Hub: GA
• Recommendations: EAP
• Prompt-Based Reporting: Available through the EAP starting the second half of 2025
• Custom QA: GA starting June 2025

One More Thing…

The final announcement from Zendesk is a new product called Zendesk Employee Service Suite. It uses pre-trained AI to help those in IT, human resources, and other support teams provide rapid responses to employee tickets. The Employee Service Suite features a service catalog, pre-built human resource information system (HRIS) integration, and an agent workspace.

“The new employee service suite is easy to implement and scale across departments, ensuring quick time to value and a low total cost of ownership,” Eggemeier said in a prepared statement. “Zendesk future-proofs employee service with an easily integrable, adaptable, and customizable solution that empowers organizations to navigate a rapidly evolving workplace while leveraging the power of Agentic AI.”

Featured Image: The Zendesk logo on display at the company's Relate conference on March 25, 2025. Photo credit: Ken Yeung