Summary: ◀▼
Automated resolutions are the usage metric for AI agents and the basis for billing. You’re charged only when a customer’s issue is resolved without human help, and each resolution is verified by an LLM. The article explains how resolutions are counted across messaging, email, web form, and legacy article-recommendation channels, plus plan allotments, purchased add-ons, and ways to avoid overage charges.
Automated resolutions are the unit of measurement used for calculating and billing your account for AI agent usage and legacy AI agent usage.
Paying per automated resolution means that you pay only for customer requests that were successfully resolved by the AI agent, without any escalation to a human agent. Conversations flagged as resolved are also verified by a large language model (LLM), ensuring its accuracy and delivering a true automation rate.
This article contains the following topics:
Understanding the default allocation of automated resolutions per plan
All Zendesk Suite and Support plans include a baseline number of automated resolutions based on your plan type. You can increase the number of automated resolutions in your account or pause AI agent functionality when you reach your limit to prevent overage. You can monitor automated resolution usage.
If you want to ensure your account isn't consuming any automated resolutions, you can turn off all capabilities related to AI agents.
Default automated resolution allocation per plan
Zendesk Suite and Support plans include a number of automated resolutions, which varies by plan type. If you need more automated resolutions, you can add them to your account. Your allocation of automated resolutions expires annually or at the end of your subscription term (if less than a year) unless you are on a non-standard subscription term.
The following table shows the number of automated resolutions included in each plan. Accounts on all plans have a maximum of 10,000 allocated automated resolutions per year. Light agents are not included in the default allocation calculation.
| Plan | Zendesk Suite | Support (standalone) |
|---|---|---|
| Enterprise | 15 automated resolutions/agent/month | 15 automated resolutions/agent/month |
|
Professional Growth |
10 automated resolutions/agent/month | 10 automated resolutions/agent/month |
| Team | 5 automated resolutions/agent/month | 5 automated resolutions/agent/month |
Avoiding automated resolution overages
You can avoid automated resolution overages by increasing the number of automated resolutions in your account or pausing AI agent functionality when you reach your limit.
You can add to your automated resolution allotment by purchasing 100 or more automated resolutions ahead of time. Also called committed usage, this approach allows you to raise the maximum number of automated resolutions available in advance. Committed usage provides a better per-resolution cost than overage billing (or pay-as-you-go).
If you choose an overage billing (or pay-as-you-go) approach to automated resolutions management, the cost per automated resolution over your allotment will be greater than the cost per automated resolution when you buy a fixed number to increase your allotment. You can prevent overage billing by limiting the number of automated resolutions available in your account each month. Overage is billed monthly, regardless of your subscription terms.
You can choose how your account responds when you reach your automated resolution limit, either allowing overage or pausing AI agent functionality if you'd like.
Understanding how automated resolutions are measured for AI agents
Automated resolutions apply to both messaging and email AI agents, and are calculated using the following logic:
- The conversation included at least one identified use case (other than a system reply) and a provided response, or else a sufficient knowledge answer provided by agentic reasoning or the Generative replies / uGPT reply.
A verification process is performed by an LLM to evaluate the text of the conversation to ensure that the customer’s request was actually satisfactorily resolved without human-agent intervention. Conversations that don’t pass this verification are not considered verified and do not consume an automated resolution.
Automated resolutions are evaluated at the end of the conversation session duration, which is by default set to 72 hours after the first message.
Only conversations that were not escalated are evaluated for automated resolutions.
Understanding how automated resolutions are measured for legacy AI agents functionality (legacy)
An automated resolution is counted when a customer’s issue is successfully resolved without live-agent intervention. Automated resolutions are counted per conversation rather than per user. A single user’s visits over multiple channels, browsers, or devices will be considered separate interactions.
There are differences in how and when an issue is deemed resolved, depending on the channel used. This section explains these differences so you can better understand your usage.
Automated resolutions in essential AI agents and legacy AI agent functionality (legacy)
If you're using legacy AI agents - Essential and legacy AI agent functionality (including Bot builder and zero-training agents), automated resolutions are calculated for AI agents messaging, email, and web form channels. No automated resolutions are counted when you test your AI agent using the testing features in Admin Center.
Messaging (including Web Widget, mobile SDK, and third-party messaging channels such as Facebook Messenger, X Direct Message, Instagram Direct, LINE, or WhatsApp). An automated resolution is counted after 72 hours of inactivity if AI evaluation has confirmed that the AI agent’s response was relevant and the last interaction was one of the following:
- The end user provided positive feedback (“Yes, problem solved”) or provided no feedback
- The AI agent shared help center articles in response to the end user’s
request:
- If articles were shared using Article Recommendation or an article step in an answer flow, the end user must have clicked on at least one article link.
- If articles were shared in a generative reply, the end user doesn’t have to click on an article for an automated resolution to be counted.
- The end user reached the final step in an answer flow.
The end user is anyone who makes a request and interacts with the AI agent.
Email and web form. An automated resolution is counted after 72 hours of inactivity if all of the following are true:
- The AI agent provided a generative reply based on the end user’s question.
- The end user provided positive feedback or no feedback.
- No human agent responded to the ticket created by the end user’s request.
- The AI evaluation process confirmed that the AI agent’s response was relevant.
Automated resolutions in legacy autoreplies with articles (legacy)
If you’re using legacy autoreplies with articles , an automated resolution is counted in either of the following situations:
- An article is suggested to the user, and the
user clicks “Yes, close my request” in:
- A feedback prompt on the article page opened via the email notification or web form.
- A feedback prompt embedded in the email notification or web form.
- An article is suggested to the user, and within
72 hours, the user clicks on the suggested article
link and
- The ticket status is marked solved before any user or public agent reply or
- No user or public agent reply has been added to the ticket.
When either of the conditions above is satisfied, then the ai_agent_automated_resolution tag is added to the ticket 72 hours after the last user interaction as long as the ticket is not yet closed.
Automated resolutions in Web Widget (Classic)
If you’re using Web Widget (Classic) to deliver article recommendations to your end users, an automated resolution is counted when the AI agent has suggested at least one article via Article Recommendations and the end user has clicked on at least one article preview link or provided positive feedback “Yes, problem solved”.
When this happens, a ticket is created for tracking purposes with a requester called End user, no assignee, and the ai_agent_automated_resolution tag. The End user profile cannot be edited or deleted.
The conversation is considered unresolved if:
- The user initiated a live chat
- The user submitted a contact form
- The user requested a callback
- The user provided negative feedback (“No, I still need help”)
- The AI agent (referred to as a “bot” in the flowchart below) didn’t understand the request
In the Web Widget (Classic), a session ends after 72 hours of inactivity or when the end user closes the browser or tab.
Actions that don’t contribute to automated resolution calculations (legacy)
Also, actions that link to another flow and automated resolutions in sandbox environments do not count against your allocation of automated resolutions.