Back in October 2025, I interviewed Tom Eggemeier, CEO of Zendesk, and during our chat, Eggemeier introduced me to the idea of a ‘service dividend’.
This idea holds that, while some companies are focused on pocketing the cost savings that come from applying AI to the bottom line, the most innovative brands in the world are starting to view those savings as a service dividend and are reinvesting them into their service experience, believing it will create a loyalty loop.
At the time, I thought it was an elegant way to describe both the choice facing service and experience leaders and the mindset shift underway, which spoke to the realities of getting the most out of CX AI investments and how to use them to drive better customer, employee, and business outcomes.
Since then, I’ve been thinking a lot about the service dividend idea, particularly how service and support leaders are using the capacity freed up by their CX investments, and how automation is reshaping the nature and shape of work in the service and support space.
Zendesk Relate: Setting the Scene
With that in mind, I was excited to be invited to Zendesk Relate, Zendesk’s annual customer event, held in Denver a couple of weeks ago. I thought it would be a great opportunity to catch up on all things Zendesk and to explore how many service and support leaders are embracing the service dividend idea.
As with every CX technology customer event, there were a lot of announcements.
To catch you up, here’s an overview of the key announcements that stood out for me:
- Zendesk announced its vision for the Autonomous Service Workforce, in which specialised AI agents will work seamlessly alongside human experts as a unified team, powered by its Resolution Platform, announced last year, and enhanced by the inclusion of a “Resolution Learning Loop” that enables the system to learn and improve continuously from every interaction.
- Outcome-Based Pricing, where businesses will only pay for interactions that are verifiably resolved end-to-end by AI, completely excluding spam or routine unhelpful exchanges.
- Expanded Zendesk AI Agents: By leveraging its recent acquisition of Forethought, Zendesk AI Agents now operate seamlessly across email, messaging, voice, and even external third-party platforms (accelerated by the recent Forethought acquisition and the introduction of Zendesk’s MCP Server). Moreover, Voice AI Agents now support over 60 languages and can switch languages mid-conversation without losing context.
A suite of specialized AI assistants to help your human team members:
- Agent Copilot: By connecting to internal and external sources, this AI assistant is designed to work out of the box and take action on at least 30% of tickets from day one.
- Admin Copilot: This assistant is designed to monitor the health of your Zendesk instance, flag operational bottlenecks, and recommend or apply workflow fixes in real time.
- Knowledge Copilot: This co-pilot autonomously analyses customer conversations in real time to identify knowledge gaps, draft new articles, and flag outdated content.
- Analyst Copilot: Allows leaders to interrogate their data in natural language, helping service and support leaders uncover the “why” behind trends and gain actionable insights.
Reading through that summary, it’s clear they have been hard at work, and their efforts have been amplified and accelerated by some excellent acquisitions (for example, HyperArc and Forethought).
Moreover, based on the customer conversations I had on the conference floor, the announcements were well received, and there was a sense of excitement at the possibilities on offer.
The Service Dividend in Practice
Amid all those conversations and announcements, I had the chance to catch up with Eggemeier again to revisit the service dividend idea.
I specifically wanted to ask him about a comment he’d made during a briefing, in which he noted that Zendesk, using their own tools, have been able to automate 60% of their Tier 1 and Tier 2 service enquiries, freeing up a huge amount of capacity.
According to Eggemeier, Zendesk has achieved this level of automation because its AI agents now not only surface knowledge base articles but also execute complex “agentic actions” to resolve issues such as billing and instance management.
He notes that achieving this level of automation has also led to a 20% improvement in their customer satisfaction (CSAT) scores.
This high rate of automation has freed up a significant amount of resources and, by embracing their own ‘service dividend’, has allowed Zendesk to:
- Create specialized, higher-value roles: As a result of automation, Zendesk has been able to shift employees away from routine Level 1 and 2 tasks and into advanced roles such as forward-deployed engineers, automation engineers, AI experts, and service architects.
- Provide direct customer consulting: These newly transitioned technical experts now work directly alongside Zendesk’s customers, helping them redesign their workflows, adopt best practices, and improve their external AI automation rates.
- Tackle long-standing product backlogs: The automation lowered the barrier to entry for coding and support, enabling Zendesk to act on low-hanging fruit that had never previously made it to the top of the priority list. For instance, Eggemeier himself was able to quickly build and ship a requested hover-bar feature for custom ticket fields that had been in the backlog for four years.
- Identify gaps and improve self-service: Using insights from their AI copilots, Zendesk identified specific gaps in their product documentation. The freed-up capacity enabled the team to proactively rewrite this documentation so customers and AI agents could self-serve more effectively.
- Launch a robust community program: The time and resources saved through AI automation finally gave Zendesk the bandwidth to launch a much more comprehensive community program, a project that had been on their wish list for a long time.
This is Zendesk’s service dividend in action.
From Technology Vendor to Transformational Partner
Now, what’s intriguing about Zendesk’s approach is how they are pushing forward-deployed engineers and other technical experts closer to customers to help them be successful.
This makes sense given that recent research suggests that many brands are still struggling to scale and generate value from their AI investments.
What’s different is that while they are figuring things out as they go, they’re also thinking about how to share their learnings with their customers.
According to John Kelleher, Vice President UKI/MEA at Zendesk, however, this is not a standalone tactic and is starting to take shape as a new engagement model.
Recognising that companies often attempt to buy AI technology without a clear strategy, Kelleher worked with an external consultant to create an engagement model that focuses on five pillars: setting a strategic vision, organizing data architecture, defining the business case, establishing AI governance, and “technical validation” and on-the-ground execution to help businesses transform their service operations.
This is helping to elevate Zendesk from a standard technology vendor into a true “transformational partner” that co-creates solutions alongside its clients.
This feels like a smart move, given the impact automation is having on the nature of work in the service and support space, and the fact that the future of work is being written as we speak – and personally speaking, I am looking forward to seeing what transformative results Zendesk and its clients can deliver by embracing this new type of vendor-client relationship.
