Help Center

Docs

AI Customer Support Agent Add-on

What the agent costs, what counts as a conversation, and what happens at the limit

What the add-on is

A customer-facing agent that answers from your verified articles, bought per agent.

The AI Customer Support Agent is an embeddable agent you put on your website or help center. It answers customer questions from your published, verified knowledge base articles, shows the customer which article the answer came from, and hands the conversation to a person when it is unsure. It is a paid add-on on the Ultra plan, bought and billed per agent, and it is not included in the plan price. The autonomous ticket-resolution AI, which drafts and resolves tickets for your own team, is a different product with its own rules. See AI Autonomy Controls for that one. For the setup walkthrough, see Set up your AI Customer Support Agent.

What it costs

Per agent, monthly or annually, in US dollars.

  • $59.99 per month, per agent, or $599.90 per year, per agent.
  • 500 conversations a month are included with each agent.
  • Run more than one agent and each one is billed separately and carries its own included 500.
  • Billing is in US dollars, and the add-on is available on Ultra plans only.

What counts as a conversation

One conversation is counted the first time the agent answers.

A conversation is counted at the first answer. Everything before that point is free: a customer opening the widget, browsing your help center, or searching your articles never counts. Once the agent has answered, the rest of that conversation is covered by the same count, no matter how many follow-up questions the customer asks in it. A conversation the agent hands to a person without answering is not counted either.

Included conversations and extra conversation credits

The monthly allowance resets. Purchased credits persist.

Each agent's 500 included conversations reset at the start of every month, and unused ones do not carry over. If you need more in a given month, you can buy extra conversation credits:

  • 1,000 conversations for $80
  • 2,000 conversations for $160

Purchased credits do not reset with the month. They stay on your account until they are used or until they expire, 12 calendar months after purchase. Conversations are drawn in a fixed order: the month's included allowance first, then purchased credits, oldest pack first.

Conversation credits are their own balance
Conversation credits are separate from Extra AI Credits and from your organization's monthly AI budget. Buying one never adds to the other, and agent conversations never draw down your AI budget. Each balance is shown on its own.

What happens when an agent runs out

The agent stops answering and falls back to article search.

When an agent has used its included conversations for the month and has no conversation credits left, it stops answering. The widget stays up and falls back to article search, so a customer can still find and read your published articles, still open a ticket, and still ask for a person. Nothing is charged automatically to keep the agent answering. Add conversation credits and it resumes straight away; otherwise it resumes on its own when the included allowance resets at the start of your next month.

The 14-day agent trial

One trial per organization, for a $1.99 deposit credited to the first month.

You can try one agent for 14 days for a $1.99 deposit. If you keep the agent, the deposit is credited to your first month, so the trial costs nothing extra. The trial is limited to one per organization. This deposit is separate from the $2.99 deposit on a Laureo plan trial, and paying one does not pay the other.

Reading transcripts

Every conversation is kept so you can read what the agent said.

Every conversation the agent handles is stored with its full transcript: what the customer asked, what the agent answered, which articles it cited, and whether it handed off to a person. Read them from the AI Agent tab in the Service hub. Transcripts are the fastest way to find the questions your knowledge base does not answer yet, so it is worth reading the handed-off ones regularly and turning the gaps into articles.

Instructions and guardrails

Instructions steer the agent. The output check is what stops a reply.

You can give the agent written instructions covering tone, length, and how you want it to handle particular topics. Treat those instructions as guidance, not a guarantee: written rules steer how the model writes, but they cannot by themselves make a topic impossible to raise.

What actually stops a reply is the output check that runs on every answer before it reaches a customer. It reads the drafted answer and hands the conversation to a person instead of sending it when the answer would quote a price, promise a refund, or accept liability, and when the subject falls inside the safety floor. That floor always hands off and cannot be turned off or lowered by an instruction.

A handoff is not a failure
An agent that hands off is working. Every handoff opens a real ticket on the customer record with the whole conversation attached, so your team picks up with full context. A handed-off conversation still counts only if the agent had already answered in it.