> ## Documentation Index
> Fetch the complete documentation index at: https://wiredesk.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Spend cap

> The hard monthly limit on what you pay past your plan: the $50 default, what stops at the limit, and how usage is counted.

The spend cap is a hard monthly limit on what you pay **beyond** your plan's included usage. The plan fee and the included usage never count against it. When usage past your plan reaches the cap, the agent stops answering billable conversations until you raise the cap or the month rolls over.

It exists so a misconfigured agent, or a bot flooding your widget, cannot run up a bill you did not agree to.

## The default is \$50

Every workspace starts with a $50 cap. That is a pilot number, not a production one. Leaving it at $50 is the most common way a busy first month pauses the agent. Set it to the overage your finance team would approve without a second conversation. Then check it against a real month's spend.

If you go past your allowance most months, look at the next plan up before raising the cap again. Its larger allowance often costs less than the overage. See [billing](/docs/account/billing).

## How the limit works on each plan

| Plan                            | When the agent stops answering                                                                                              |
| ------------------------------- | --------------------------------------------------------------------------------------------------------------------------- |
| Free, no card                   | When the Free allowance (500 replies or 50 voice minutes) is used up. Nothing past it is billed, so the cap does not apply. |
| Free with a card, Growth, Scale | When usage past the allowance reaches your cap.                                                                             |
| Enterprise                      | When usage reaches your cap. Enterprise has no included allowance. Volume rates apply from the first reply.                 |

## Change the cap

Only the owner can change it.

<Steps>
  <Step title="Open Billing">
    On the **Beyond your plan** card, select **Update limit**. On the dashboard, you can also select the dollar amount on the same card.
  </Step>

  <Step title="Enter the new limit">
    Anything from $10 to $10,000 a month.
  </Step>

  <Step title="Save">
    The new limit applies straight away. If the agent had stopped, it answers again as soon as the new limit is above what you have already spent.
  </Step>
</Steps>

Every change is recorded in **Settings → Activity**.

## Warnings before you hit it

WireDesk notifies the owner, by email unless you change it, at most twice a month:

* **80% of your monthly limit** — sent once a month, while there is still time to raise it.
* **Monthly limit reached** — when conversations pause.

On Free without a card, the same two notices count against the Free allowance instead. You can send them by text as well, or switch them off, in [notifications](/docs/workspace/notifications).

## What happens at the limit

Nothing is lost. Incoming messages are still stored, and you can read them in **Conversations**. The agent just does not answer them.

| Where                     | What the customer or tester sees                                                                                                            |
| ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------- |
| Website chat              | Each message gets this reply instead of an answer: "Sorry — this assistant is temporarily unavailable. Please reach the business directly." |
| Email                     | The email is stored, and no reply is sent.                                                                                                  |
| Voice test in the browser | The test does not start. The message reads "monthly spend cap reached — raise it on the dashboard to keep testing".                         |

**Billing** and the dashboard both show that the agent has stopped. Replies your team writes from the dashboard are not affected, because they are not billed.

## How usage is counted

Each billable unit is recorded as a usage event of one of three types:

| Usage event     | Recorded when                                                               | Counts as                                   |
| --------------- | --------------------------------------------------------------------------- | ------------------------------------------- |
| `chat_message`  | The agent sends a reply in website chat, including the Playground test chat | \$0.02                                      |
| `email_message` | The agent writes a reply to an email                                        | \$0.02                                      |
| `voice_seconds` | A phone call ends, or while a browser voice test runs                       | \$0.20 per started minute, per conversation |

A few details:

* **An email reply** is counted when the agent writes it, not when it is delivered.
* **A phone call** is counted when it ends, rounded up to the next whole minute.
* **A browser voice test** of your agent charges its first minute when it starts. After that, each further started minute is charged while the page stays connected. A test runs for up to 15 minutes, then ends with "That's the longest a session can run. Start another to keep going." Only the owner and admins can start one.
* **The demo agent** at `/talk` is not charged to your workspace. Each account gets 10 minutes of it a day, up to 5 minutes per session.
* **Not counted:** teammates, agents, knowledge sources, and replies your team writes themselves.

The month-to-date figures on **Billing** add up these events. Use that page when you want the exact numbers.

## Related

* [Billing](/docs/account/billing)
* [Plans and limits](/docs/reference/plans-and-limits)
* [Notifications](/docs/workspace/notifications)
* [Test your agent](/docs/agents/test-your-agent)
