Agent Feed is your morning inbox for automated agent runs. Any time an agent runs on a schedule or a trigger, the result shows up here — one card per run, grouped by day, across every deal your agents are watching. Instead of clicking through deals one by one to see what changed overnight, you open Agent Feed once, triage what matters, approve or reject the actions your agents want to take, and move on.

This article covers how to get around the feed, what each card tells you, how to act on agent actions directly from a card, and how to keep your queue clean. For the full lifecycle of an agent action (statuses, failures, permissions), see Manage agent actions. For setting up the automations that feed this page, see Set up agent triggers.

How to access

In the web app

Open Pod.
In the left navigation, click Agent Feed.
If Pod shows a homepage banner for new results, you can also click Agent feed there.

In the Chrome Extension

Open the Pod Chrome Extension while you work in your CRM.
Open the menu from the top-left corner of the extension panel.
Select Agent Feed.

What a feed card shows

Each card is one agent run. It includes:

The agent name that produced the result.
A Personal or Org badge so you can tell if the result came from your own trigger or a shared workspace trigger.
A link to the related deal (the record the agent ran against) — clicking the record name opens that deal.
A short result summary from the agent.
A relative timestamp (e.g. 12m, 2h, 1d) so you can spot fresh activity.
A blue unread dot on the right until you've opened the card.
If the agent proposed any actions, an actions section at the bottom of the card (approve/reject controls or a Review actions button — see below).

Clicking anywhere on the card opens the related deal's AI Agent view, jumping you straight into the run's conversation thread.

Filter and date groups

The header has a single filter dropdown with three options:

Read & Unread (default) — everything in your active feed.
Unread — only results you haven't opened yet.
Archived — results you've filed away.

Below the filter, results are grouped by date: Today, Yesterday, This Week, and Earlier. Each group header shows a count and can be collapsed with the chevron on the right.

Archived results live under the Archived filter — they are not on a separate tab, and they don't clutter your Read & Unread view.

Act on agent actions right from the feed

If an agent proposed write actions (like sending an email, pushing a CRM update, or creating a task), those actions show up on the card and you can approve or reject them without opening the deal.

One or two pending actions

When a card has one or two pending actions, each appears as its own row inside the card with the action label, an approve button, and a reject button. Both buttons use Pod's hold-to-confirm interaction — press and hold for half a second to commit. This prevents accidental approvals of write operations.

Three or more pending actions

When a card has more than two pending actions, Pod collapses them into a single row that shows:

A yellow badge: <N> actions pending.
A Review actions button that opens the deal's AI Agent view so you can approve or reject each one in the full thread.

Stale actions

Agent actions that sit pending for more than 24 hours are marked stale. You'll see a Stale — revalidation may fail caption on the card (or a small stale badge in the collapsed view). Stale actions can still be approved or rejected, but the target system may reject the write because the underlying data has moved on.

Resolved actions

Once every action on a card has been approved or rejected, the actions section collapses into a green badge: <N> actions resolved. That's your signal you're done with the card.

For the full lifecycle of an agent action — every status, every failure reason, and who can approve what — see Manage agent actions.

Actions on feed cards work the same way in the Chrome Extension and the Pod web app.

Triage a single result

Hover (or focus) a card in the Read & Unread or Unread view to reveal three quick actions in the top-right:

Trigger reason (⚡) — tooltip that explains why this run fired (useful when you're trying to figure out whether a trigger is too noisy).
Mark read / Mark unread — flip the unread state on a single card.
Archive — move the card to Archived.

Clicking the card body (or pressing Enter while it's focused) opens the related deal's AI Agent view at the exact run.

Bulk actions

The ⋯ menu in the header offers two bulk controls that apply to the current active feed:

Mark all as read — marks every unread active result as read.
Archive unread — archives every unread active result in one move.

Use these when you want to get the feed back to zero without opening every card.

Review a teammate's feed

If you manage other users in your workspace, you'll see a user selector at the top of the feed (next to the filter dropdown). Switch users to review the runs that landed in their feed. Same filter, same cards, same approve/reject behavior — just scoped to the selected user. Only people with managed access see the selector.

Jump from the feed to your automations

The bot icon in the feed header (Manage automations tooltip) routes to Settings → Automations, where you can pause, edit, or remove the trigger that produced a noisy card. Use this whenever the result on a card makes you want to change the trigger.

For the full walkthrough of setting up and tuning triggers, see Set up agent triggers.

Empty states

If you haven't set up any automations yet, the feed shows a No agent results yet empty state with a single call to action:

In the web app: Go to Automations (opens Settings → Automations).
In the Chrome Extension: Open Automations (opens Pod in a new browser tab at Settings → Automations, since Settings is web-app only).

When to use Agent Feed vs. other AI agent pages

Use Agent Feed when you want to review what your automated agents have produced and act on the findings that matter.

Go to these related pages when your goal is different:

Use Launch and use AI agents when you want to run or continue an agent on a live deal right now.
Use Manage agent actions when you want the full reference for approving, rejecting, and troubleshooting agent actions.
Use Set up agent triggers when you want to change what's running automatically.
Use Build and manage agents when you want to build, organize, or import the agents themselves.
Use Build a new agent when you want a deep walkthrough of writing a strong prompt and response format.
Use the AI Agent Best Practices Guide when you want help making your agents better over time.