Your Zapier agent paused and asked a question
It looks like a broken build. It isn't — it's a per-run limit doing exactly what it's designed to do.
The short version
Your agent has not broken and you have not misconfigured it. On Zapier's free plan a single run can use up to 10 activities. When a run reaches that ceiling, the agent stops where it is and asks you for input instead of carrying on to a finished answer.
It looks like a failure because it arrives in the same place a result was supposed to. It is a limit doing exactly what it was designed to do.
What was being counted
An activity is the agent reaching outside itself — running a web search, fetching a page, using a trigger, running an action, answering from a knowledge source. Talking to the agent is not counted; going and getting something is.
So a run that searches, then reads, then reads again, then reads again is working through that allowance of 10 as it goes. A task defined broadly enough will reach the end of it before it reaches the end of the job.
What to do about it
Two routes, and they are genuinely different in what they teach you.
Answer the question it asked. The agent stopped because it wanted input; giving it input is the direct response. We have not measured what continuing costs against your allowance, so we will not pretend to know.
Or narrow the job and run it again. Tighten the instructions so the agent is doing less reaching — name how many sources it should check, or make the topic narrower — and start over.
The second is the more useful habit, and not only because of the allowance. Read what came back, work out what was too loose, change one thing, run it again. That loop is most of the skill of working with agents, and hitting a limit is just an unusually blunt prompt to practise it.
Stopping it happening again
The limit is on what a single run reaches for, so the lever is how broadly you have told the agent to search. Instructions that say exactly how many sources to check give the agent a stopping rule. Instructions that say "research this thoroughly" do not, and the agent has no way to know that thoroughness has a budget.
We are not going to hand you a specific number of sources as a safe setting, because the cost of any given run depends on what the agent chooses to do inside it, and we have not measured that. Start conservative, watch what happens, adjust.
One thing worth checking
The per-run ceiling is 10 on the free plan and 40 on Pro. If your account is less than a fortnight old you may be inside Zapier's 14-day Pro trial, and we have not tested whether that changes which ceiling applies to agent runs. If the behaviour you are seeing does not match this article, that is the first thing we would look at.
The other free-plan limits — the monthly allowance, what resets when, and why test runs are more expensive than people expect — are in our write-up of the free tier.
Source and date
Checked on 23 August 2026 against Zapier's usage documentation. These limits change — if you are reading this long after that date, check before relying on it.
Build your first agent in 30 minutes
One guided session, and you finish with a working agent you can run and adapt. No coding experience required.