Zendesk’s pricing has come a long way—from standalone products and add-ons to a unified Suite, and now toward AI-powered flexibility.

In the early days, you had Zendesk Support. Guide, Chat, Talk, and Explore were sold as separate products—each with its own tier: Team, Professional, or Enterprise. You could add Collaboration to get light agents, or Multibrand to unlock more flexibility.

Then in 2021, Zendesk bundled it all into the Suite. Cleaner tiers, one subscription. No more mix-and-match licensing. Just Suite Team, Suite Growth, Suite Professional, and Suite Enterprise.

Fast forward to today, and we’re back to add-ons. You’ve got Suite at the core. But you might also need Agent Copilot, the ADPP platform add-on, or the QA and WFM tools to get the full picture.

Despite these shifts, the underlying pricing model stayed the same: you pay per agent. Add more agents, and your cost scales linearly. In return, you get unlimited usage—within fair storage limits—and full access to the platform features you’ve paid for.

This model works well when humans do the work, and Zendesk is the tool that helps them do it faster.

But Then AI Showed Up

The agent seat model made sense in a world where most tickets were handled by people. More customers meant more tickets, which meant more agents—and higher license costs. It scaled with your business.

But AI introduces a second model.

It started quietly, with Answer Bot and Sunshine Conversations, which priced based on monthly active users. Then came Ultimate, which brought a conversation-based fee model. Zendesk acquired it, evolved it, and eventually launched AI Agents—priced not by seats, but by interactions.

That shift makes sense too: more customers = more conversations = more AI usage.

But here’s the rub: when an AI Agent fails to resolve a request and escalates it to a human, you’re charged twice. Once for the AI interaction, once again for the agent who takes over.

A Fairer Model: Automatic Resolutions

This is where Zendesk’s automatic resolution pricing comes in. You don’t pay for every AI conversation—only the ones that result in a resolution.

If AI handles the full conversation and resolves the issue? You pay. If it escalates to a human? That’s covered under your existing agent license.

This is a fairer model, because it ties cost to value. It also aligns incentives: better automation means fewer escalations, which means lower total cost of support.

There's an argument to be made that we used to get chatbots for free with Zendesk and they now all of the sudden started charging for that usage. But in my opinion, the generative agents we get today are in no way comparable to the chatbots of old when it comes to their automation capability. You could almost agree it's a different product altogether.

We only get paid for our AI agent if we solve our customer’s customer problem (…) You’d have flexibility between human and automated resolutions… and that aligns our business model to our customers’ model better.
Zendesk CEO Tom Eggemeier —Punk CX Podcast, April 2025

And now with added complexity

Here’s the problem.

As Zendesk’s AI offering grows, so does its product complexity. It’s no longer true that buying Zendesk Suite gives you everything.

To get the full AI experience today, you need:

• Suite Professional or higher
• AI Agents Advanced
• Agent Copilot
• Zendesk QA

That’s a big ask—especially when summarization, writing tools, and sentiment detection are rapidly becoming table stakes. Zendesk addressed part of this recently with the addition of generative replies for conversations and email and the new Quick Answer for Agents and Help Center.

Not every company knows how much of their ticket volume is automatable. AI capabilities evolve. Ratios shift. You start with 20% of tickets resolved by AI, then suddenly you’re at 50%. But how do you budget for that?

Zendesk seems to agree. In a recent pilot involving 20–30 customers, the company is testing a flexible pricing model that allows teams to trade between human and AI agent resolutions. Instead of paying for seats, customers buy a bucket of resolutions—with AI resolutions priced higher than human ones, given their automation value.

So we've got a system where people can go trade between human agents and Al agents are licensing between automated resolutions and right now seats, so that they're invested protected and they can go scale up or down, more human agents, more agents budget, and then change the levels. (...)
We're piloting this right now with about 20 or 30 of our customers in the globe. The idea would be that you would have a bucket of resolutions. Quite frankly, human resolutions are going to be priced less than AI edge at resolutions because you have the human labor attached to that.
Zendesk CEO Tom Eggemeier —Punk CX Podcast, April 2025

Future: The AI-Powered Shift

Zendesk’s own projections make the direction unmistakable:

• 100% of service interactions will involve AI
• 80% will be fully resolved by AI
• 20% will still require a human agent

Every interaction—whether through a chatbot, Copilot reply, or AI-powered knowledge lookup—will be shaped by AI. Most will be handled without human involvement. And only the most complex or high-value issues will escalate to a person.

This prediction reflects a deeper shift. Zendesk is moving from a seat-based model to one based on resolutions. What matters isn’t how many agents you have, but how many tickets get solved—and how.

A Cleaner Approach

Here’s a vision: what if Zendesk Suite just included everything (again)?

Imagine a new Suite that bundles the features of AI Agent Advanced and Agent Copilot into one platform. You get:

• Core platform with Agent workspace, Help Center, Analytics and the rest of the Admin Center
• Agentic AI with procedures and generative replies
• Intelligent Triage with use case, sentiment, and intent detection
• Writing Tools with suggested replies, writing tools, and summarization
• Agent Copilot with Auto-assist, actions and procedures

All part of your subscription. But instead of paying for feature access via agent seats, you pay for what gets used:

• Automatic resolution fee for tickets resolved by AI (e.g. generative replies or autonomous procedures). Currently this is roughly $1 per resolution.
• Agent resolution fee when AI helps a human agent (e.g. Copilot fires a procedure, suggests a reply, or summarizes a thread. This fee would be lower in cost, since we still need to take wages and related costs into account.

Now, this Zendesk article puts the average amount of tickets per agent at around 100. And this other source puts it at 400. So let's average that out to 250 tickets per month assuming that tickets become more complex as AI takes over the "easy" ones. Current agent cost with Agent Copilot for Zendesk Suite Professional is around $150/month list price, which means you're currently paying around 50-60cts per resolution already. So let's just say, Agent Resolutions are 1/2 the cost of AI Agent resolutions in this model.

Regardless of the actual price, by shifting from agent seats towards resolution fees, this new model makes the platform:

• Predictable – You pay for outcomes, not features.
• Aligned – AI is incentivized to succeed. Escalations don’t get double-billed.
• Unified – No more mix of add-ons and tiers to unlock core functionality.

A rough way to estimate the required amount of resolutions is to dive into explore and look at your ticket replies. Any single-reply tickets can roughly be equated to AI resolutions resolved via Help Center articles or procedures.

Multi-reply tickets can roughly be equated to Agent resolutions, even though some will naturally shift toward AI Agent procedures over time with the help of API integrations.

Final Thought

Zendesk’s move toward AI is bold—but its packaging needs to catch up. A simplified model where AI features are built in, and pricing reflects usage and resolution, is the next logical step.

It’s better for customers, clearer for finance teams, and more scalable for Zendesk itself.

The future isn’t per seat. It’s per resolution.
And we’re already halfway there.