Flagging an Agent Message

When your Agent gives a reply that misses the mark, you can flag it right from the Conversation. A flag records what went wrong, how serious it was, and what the Agent should have said instead. It's the simplest way to turn a bad answer into feedback the TrustYou team can act on, and to keep track of the replies your own team wants to revisit.

What Flagging Does

Flagging marks a single Agent reply as a problem and attaches your feedback to it. Think of it as leaving a note on that exact message: what was wrong, how bad it was, and what a good answer would have looked like. The flag stays on the message so you and your teammates can find it later, and it feeds back to TrustYou to help improve how your Agent responds.

You flag individual Agent replies, not whole Conversations. Guest messages, staff messages, and system messages can't be flagged, only the answers your Agent gave.

How to Flag a Message

1. Open the Conversations page and click the Conversation you want to review. See Browsing and Filtering Conversations.
2. Find the Agent reply you want to flag. Click the flag icon in the top corner of the message.
3. The Flag this message dialog opens. Fill in the details (all four are required):
   • Severity -- how serious the problem is: Low, Medium, High, or Critical.
   • Category -- what kind of problem it is. Pick one or more:
     • Wrong information
     • Missing information
     • Tone of voice
     • Factual error
     • Unhelpful
     • Off topic
     • Other
   • Reason -- a short note explaining what went wrong with this message.
   • Desired outcome -- what the Agent should have done or said instead.
4. Click Submit complaint. You'll see a confirmation, and the flag appears right under the message.

Where Your Flags Show Up

Once you've flagged a message, the feedback is visible in a few places so nothing gets lost:

• On the message -- a card appears right below the flagged reply. Collapsed, it shows the severity and categories. Expand it to read the full reason, the desired outcome, and when it was filed.
• In the Conversation header -- a pill shows how many messages in the Conversation are flagged.
• In the Conversation list -- a flag badge with a count appears on any Conversation that has flagged messages, so you can spot them while scanning.
• In the Flagged filter -- filter the whole list down to flagged Conversations. See Browsing and Filtering Conversations.

Editing or Deleting Your Flag

You can change or remove a flag you filed at any time:

• Edit -- open the flag's card and click the pencil icon, or click Edit complaint on the message. Update any of the fields and click Save changes.
• Delete -- click the trash icon on the card. You'll be asked to confirm, since deleting a flag permanently removes it and can't be undone.

Who Sees and Manages Flags

• Anyone on your team who's a member of the Organization can flag a message.
• Everyone on your team can see all the flags filed on a Conversation, so it's easy to compare notes.
• Only the person who filed a flag can edit or delete it. You'll see other people's flags too, but as read-only.

Note: You can file one flag per message. If you've already flagged a reply, the button changes to Edit complaint so you update your existing feedback instead of adding a second one. Different teammates can each file their own flag on the same message.

What Happens After You Flag

Your feedback goes to the TrustYou team, who use it to improve how your Agent answers. There's no status to track on your side, like "open" or "resolved". A flag simply stays on the message as a record of the feedback until you edit or delete it.

How We Use Your Flags

Here's the important part: we don't chase down and "fix" each flag one by one. That might sound like we're ignoring individual reports, but it's actually the opposite. It's how we make your Agent better for everyone, and we do it on purpose.

Why? Because tweaking an Agent to nail one specific answer often makes it worse somewhere else. Push it hard to fix a single reply, and it can start giving odd or overcorrected answers in other Conversations you never had a problem with. Chasing one case at a time quietly breaks two others.

So instead, we look at the bigger picture. We regularly review all the flagged messages for your Agent together, spot the patterns across them, and make corrections that lift the whole thing at once. A batch of real examples tells us far more than any single reply, and it lets us improve your Agent without knocking it off balance elsewhere.

What this means for you:

• Keep flagging. Every flag is a data point. The more real examples we have, the sharper the improvements we can make.
• Be specific. A clear reason and a concrete desired outcome make your flag genuinely useful when we review the batch.
• Don't expect a reply to each flag. No response to a single flag isn't a flag being missed. It's us doing the careful, batched work that actually makes your Agent better, rather than the quick fix that would make it worse.

Think of your flags less like support tickets and more like votes and notes that guide where your Agent goes next.

What's Next

• Browsing and Filtering Conversations -- find and review the chats worth flagging
• How the Agent Works -- understand why an answer turned out the way it did
• Knowledge Best Practices -- turn flagged replies into better knowledge so your Agent improves