Skip to main content
Wren has a fair-use limit to keep usage predictable on the platform and protect every account from runaway costs. The exact size of the limit and how often it resets are managed by PolyAI per account and may change as the platform evolves.

​
How the limit works

  • Scope: each person gets their own allowance. It isn’t a shared pool. A colleague running a long build doesn’t eat into what you have left. The size of that allowance is a setting on your account, so it’s the same for everyone on it.
  • Reset cadence: typically a regular interval (daily, weekly, monthly) chosen for your account, resetting on UTC boundaries. Your PolyAI representative can tell you which applies to your account.
If your team consistently needs more capacity than your current limit allows, contact your PolyAI representative. Limits can be reviewed and adjusted on the account.

​
What counts toward the limit

The limit tracks how heavy your interactions with Wren are: the underlying work it does to plan, reason, and apply changes on your behalf. In practice, the things that draw down your allowance fastest are:
  • Long conversations with lots of context to reload each turn.
  • Requests that touch many flows, topics, or entities at once.
  • Iterating on the same plan many times before approving it.
  • Deep analysis across hundreds of conversations.
Voice dictation and web searches also count, at a small fraction of the cost of a normal message. Requests made through the Wren API draw on the same allowance as the chat panel.

​
What happens when you hit the limit

When you’ve used your full allowance, Wren pauses until the next reset. The chat panel replaces the prompt input with a rest card telling you when your credits refresh, plus a Talk to sales link if you need more capacity. Two ways forward:
  • Wait for your allowance to refresh at the next reset.
  • Talk to sales if your team consistently needs more. Your account limit can be reviewed. Use the Talk to sales link on the rest card, or contact your PolyAI representative.
While Wren is resting:
  • The chat panel stays available, you can read past conversations and review plans.
  • Manual editing in Studio is unaffected. You can keep building flows, topics, and other configuration the normal way.

​
Other transient errors

Wren also surfaces status messages when something goes wrong outside of usage limits. These appear as a banner above the input and are temporary. Retry with the action shown. You may also see an in-chat error reply titled “I can’t process this message”. That’s a hard validation failure. Start a new chat instead of retrying. If any of these persist, contact PolyAI support with a timestamp and the conversation ID.

​
Content policy

Wren screens what goes in and what comes out against PolyAI’s content policy. A blocked message is rejected with a Content policy violation notice, and any files attached to it are discarded.
Repeated violations can result in account suspension. If you believe a message was blocked in error, contact PolyAI support rather than retrying it.

​
Tips for staying within your limit

  • Ask for a plan before big work. A single planning round is much cheaper than asking Wren to build, then redo, then redo again.
  • Scope each request to one unit of work. Smaller, focused requests reload less context and spend less.
  • Start a new chat for new work. A long chat reloads all its history every turn, even the parts that are no longer relevant.
  • Use Wren for exploration too. Read-only questions about an existing agent are cheap and don’t risk wasted build cycles.
  • Approve plans early. Long back-and-forth on a plan accumulates spend before any change is written.
See the prompting guide for patterns that produce good results in fewer turns.
Last modified on August 13, 2026