nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

How do I wait for the full response of the Rovo agent

Ivan LN

I built an agent that checks content against company guidelines and adds a comment in the page with the suggestion.

I want to plug it into an automation so it gets triggered every time a document goes "In Review"

The challenge I'm facing is that when the agent is triggered it replies with the intention of the action

I will review the document at the provided URL for compliance with our company guidelines

Please hold on while I access and review the content.

The automation treats this as the agent's response and adds it as a comment, instead of the full report that will be generated later.

I will test with giving instructions to the agent so it does not prompt the summary of the actions to come, but I was wondering if there is a more elegant solution or something that I can control in the automation.

Accepted answers

Charlie Misonne

Hi @Ivan LN

I will test with giving instructions to the agent so it does not prompt the summary of the actions to come

I think this is your only option at the moment. Let me know how you testing goes because this is an interesting case!

In automations there is no way to wait for the final answer unlike what can be done with the "send web request" action.

Alex Gallien

It would help to see your prompt to see how you're building your output. I use variations of the phrase

Return the following output without any additional commentary or formatting:

whenever I am working with an agent that interacts with automation. I also recommend having a dedicated scenario configured for automation that only returns the output you're looking for. I walk through an example of that in this workshop, the prompts for that agent can be found here: Lets Build a Rovo Agent Documentation - Triage Automation (10/15/25).

All comments

Matteo Gubellini _SoftComply_

Charlie Misonne

Hi @Matteo Gubellini _SoftComply_

It's good to know you're not the only interested in this question. Just know that community members often filter for unanswered questions. And this "+1" answer will make it look like it has been answered.

Using the like button should be enough ;-)

Charlie Misonne

Hi @Ivan LN

I will test with giving instructions to the agent so it does not prompt the summary of the actions to come

I think this is your only option at the moment. Let me know how you testing goes because this is an interesting case!

In automations there is no way to wait for the final answer unlike what can be done with the "send web request" action.

Alex Gallien

It would help to see your prompt to see how you're building your output. I use variations of the phrase

Return the following output without any additional commentary or formatting:

Ivan LN

Thanks Alex. This is what I ended up doing and it works well.

I also ended up splitting the agent in two (grammar check and tone of voice check) so each one of them has a clearer set of instructions.

Alex Gallien

Glad to hear it @Ivan LN! In theory you should be able to leverage scenarios to have one agent handle both the grammar and tone checks but no big disadvantage to having one agent handle both.

Alex Gallien

@Matteo Gubellini _SoftComply_ if you're interested in this still, looks like my answer above helped