> ## Documentation Index
> Fetch the complete documentation index at: https://docs.exerta.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Live Run Logs: Real-Time View of Every Agent Action

> Live Run Logs show every comment, DM, and chat handled by your agents in real time — including outcome, duration, step count, and any errors.

Live Run Logs give you a real-time audit trail of everything your agents do — every reply, hide, lead capture, and escalation, with full outcome details. Whether you want to verify that a specific comment was handled correctly, investigate an error, or simply watch your agents work, the log is the most direct window into your automation.

## What a log entry shows

Every entry in the log captures the full picture of a single agent action.

* **Action summary** — a plain-language description of what happened and the outcome, e.g. "Visitor about to leave · Saved" or "Question answered · Lead captured." This is the headline you scan when reviewing the log quickly.
* **Steps taken** — the sequence of actions the agent executed to reach the outcome, e.g. exit intent detected → greeted → discount offered. This shows you the agent's reasoning in order.
* **Duration** — how long the entire action took from trigger to completion. Most actions finish in under two seconds.
* **Status** — one of three states: **Success** (action completed cleanly), **Running** (action is still in progress), or **Error** (action failed and is queued for retry).
* **Time** — how long ago the action occurred, updated live as you watch the log.

## Log entry examples

| Action                            | Steps                                    | Duration       | Status     |
| --------------------------------- | ---------------------------------------- | -------------- | ---------- |
| Visitor about to leave · Saved    | Exit intent → greeted → discount offered | 1.4s · 3 steps | ✓ Success  |
| Question answered · Lead captured | Sizing question → email captured         | 0.9s · 3 steps | ✓ Success  |
| Chat in progress                  | Checkout question · answering now        | 1.3s elapsed   | ⏳ Running  |
| Post-purchase survey done         | Feedback logged for your ads             | —              | ↻ Retrying |

Every conversation, capture, and save appears in this log — nothing is processed in the background without a record here.

## Error handling

If an action fails — for example, due to an API timeout or a platform rate limit — Exerta retries it automatically after a **30-second delay**. You don't need to do anything. The log entry updates to show the retry status so you can see it's being handled.

If an action fails repeatedly after retries, it's flagged in the log with a red status badge. At that point, check whether the underlying cause is a disconnected integration or a platform issue, and reconnect the channel if needed. Persistent errors that can't be auto-resolved will also surface as a notification in your dashboard.

## Filtering logs

You can narrow the log to exactly what you're looking for using the filter bar at the top of the page:

* **Channel** — show only Facebook, Instagram, TikTok, or live chat actions
* **Status** — filter to Success, Running, or Error entries
* **Action type** — isolate replies, hides, lead captures, escalations, or sales recoveries
* **Date range** — review a specific time window within your 90-day history

Combining filters is the fastest way to investigate a specific issue — for example, filtering to Error + Instagram DMs during a specific date range.

## Using logs to improve your agents

The log isn't just an audit trail — it's the best signal you have for making your agents better. Look for these patterns:

* **Recurring errors on a specific action type** — if the same kind of action is consistently failing, check whether the connected platform credentials are still valid or whether a rule is referencing a deleted product or link.
* **Long durations on certain steps** — if some actions are taking significantly longer than others, it may indicate an external API your agent is calling is slow. Consider simplifying the step sequence.
* **Unexpected outcomes** — if a comment you expected to be hidden was replied to instead, or vice versa, open the log entry and review the step sequence. Then adjust the corresponding rule in your agent's settings or add an example to its training to sharpen that decision.

Small tuning based on log patterns compounds quickly — agents that are reviewed and adjusted weekly tend to outperform untouched ones within a month.

<Note>
  Logs are retained for 90 days on all plans.
</Note>


## Related topics

- [What Is Exerta? AI Engagement Agents for Social Ads](/introduction.md)
- [Get Started with Exerta: Live AI Agents in 5 Minutes](/quickstart.md)
- [Website Chat: AI Agent for Landing Pages and Checkout](/channels/website-chat.md)
- [Analytics Dashboard: Track Replies, Hides, and Revenue](/analytics/dashboard.md)
- [Revenue Attribution: Track Sales Back to Every Comment](/analytics/attribution.md)