The activity feed and thread timeline

    A running record of every action taken on a conversation, by people and by Chloe, with flags marking the replies worth checking.

    5 min read · Updated 7 September 2026

    The third card in the context panel is the Feed. It shows a history of all the major actions taken on this conversation, by humans and by AI.

    What the feed is

    The feed answers "who did what, and when?". Where the summary tells you what the conversation is about, the feed tells you what has been done to it.

    Its most valuable job is keeping a record of what Chloe did. When she is handling conversations on her own, the feed is how you check her work without reading every thread end to end.

    The feed and the timeline are different

    Gruvi records events in two places, and it helps to know which is which.

    The timeline
    Events written into the conversation itself, in the middle column, among the messages. Things like Handed off to a human, or a line recording that somebody marked the thread resolved or reopened it.
    The feed
    The card in the right-hand column, listing actions as a separate list, newest activity at the bottom.

    The timeline shows you an event at the moment it happened in the conversation, which is what you want while reading. The feed shows you everything in one list, which is what you want while checking.

    What each entry shows

    Every entry in the feed has the same four parts:

    Who
    Their name and photo. Chloe's entries carry a small blue AI tag, so you can tell at a glance whether a person or the assistant did it.
    When
    How long ago, such as 2w.
    What
    The action in plain words: Sent a message, Replied to the customer, Marked this thread as resolved, Reopened this thread.
    A preview
    The first line of what was written, so you rarely need to open anything.

    The two flags in the feed

    Some of Chloe's replies carry an orange badge. There are two, they mean different things, and telling them apart is the single most useful thing in this article.

    Flagged for a human
    Chloe stopped. She could not deal with this and asked for a teammate, so the conversation is now yours. You will also see a matching Handed off to a human event in the timeline.
    Flagged for review
    Chloe carried on. She answered the customer, but she was not confident in the reply and is telling you so.

    The second one is the one people miss. A reply flagged for review has already gone to the customer. If it was wrong, you need to know now.

    Scan for Flagged for review first when you check the Autopilot queue. Those are the replies where Chloe herself is unsure, so they are the best possible use of your attention. They are also the fastest to correct while the conversation is still warm.

    How to expand the feed

    The feed sits at the bottom of a column shared with two other cards, so it starts small.

    1. 1Open a thread and find the Feed card in the right-hand column.
    2. 2Click the expand icon in the card's title row.
    3. 3The feed fills the whole column, and you can read the conversation's full history without scrolling in a small box.
    4. 4Click the icon again to put the other cards back.
    5. 5The chevron beside it collapses the feed entirely, if you would rather have the sentiment and summary cards larger.

    The ? beside the title explains the card in a sentence.

    What is coming next

    The feed is being built out. Two things are visible but not yet working, so you know what they are when you see them.

    • A message box at the bottom of the panel, marked Soon. It will be team chat about a conversation, inside the conversation.
    • More agents than Chloe. The row of faces on the feed's title bar shows the teammates joining her.

    In time you will be able to ask AI to take actions from the feed directly, rather than only reading what it has already done. ---

    Was this helpful?