> ## Documentation Index > Fetch the complete documentation index at: https://learn.getodin.ai/llms.txt > Use this file to discover all available pages before exploring further. # Task Monitor > Super Admins can monitor, filter, and analyze all background tasks and automations in Odin AI. Track performance, spot issues, and optimize system reliability. The **Task Monitor** tab helps you track how Odin AI is performing behind the scenes. Every time a user runs a workflow, uploads a document, uses an AI agent, or triggers automation, it creates a **task**. These tasks are processed by the system in real time. The Task Monitor gives you a clear view into all those activities, helping you: * Understand what’s happening across the system * Spot any issues before they affect users * Improve system performance by identifying slow or failing processes If you’ve ever wondered, “What’s slowing things down?” or “Why did a task fail?” — this tab is where you’ll find the answers. *** ### Dashboard View Task Monitor *** ### Filters (Top Panel) Use filters to get results based on: * **Project ID**: Isolate task activity from a specific team or initiative. * **User ID or Email**: Audit or troubleshoot tasks run by a specific user. * **Date Range**: Defaults to the last 7 days, but can be adjusted. * **Refresh Button**: Pulls in the latest data from the task execution logs. *Tip: Always set filters before analyzing graphs — it ensures you’re only reviewing relevant data.* *** ### Top 10 Time-Consuming Tasks This bar chart lists the **tasks that are taking the most time to execute**, based on system records. * **Avg Duration (Blue Bar)**: The average time this task takes across all executions. * **Median Duration (Purple Bar)**: The middle value — helps you spot if a few slow tasks are skewing the average. This chart helps answer: * Which tasks are the most resource-intensive? * Are there outliers (e.g., one task spiking to 3x normal time)? * Should you optimize, reschedule, or break apart any specific logic? *Example: If a document extraction task consistently takes 3 minutes, but 90% of others finish in 10 seconds, you may need to revise input formats or split large files.* *** ### Overall Success Rate (Donut Chart) This chart shows the health of your system in terms of **task outcomes**: * **Success** – The task completed without errors. * **Failure** – The task encountered an error (e.g., API failure, timeout, bad data). * **Terminated** – The task was forcefully stopped before completion, either manually or by system limits. *Tip: A high failure or termination rate is a red flag. You may need to look into misconfigured agents, expired API tokens, or inefficient workflows.* *** ### Task Execution Distribution This chart shows **how often each task runs**, and how often it succeeds or fails. For each task: * Green bars = Successful runs * Red bars = Failures * Orange bars = Terminated runs It helps answer: * Which tasks are used most frequently? * Are there tasks that run often but frequently fail? * Are critical workflows being disrupted by repeated failures? *Example: A task that runs 10,000 times a week with 20% failures could be silently breaking downstream workflows — this chart helps you catch that.* *** ### Task Duration Range This graph breaks down how long each task is taking, over time. For each task, it plots: * **Average Duration** * **Median Duration** * **Max Duration** * **Min Duration** This is where you look for **spikes or volatility**. Tasks that normally run in seconds but occasionally take minutes could indicate: * Backend slowdowns * Input anomalies * Rate-limiting from third-party APIs Use this chart to identify which tasks need retry logic, parallel processing, or better input validation. *** ### What Admins Should Do Regularly | Task | Why It’s Important | | ---------------------------------------------- | ------------------------------------------------------------ | | Check the top time-consuming tasks weekly | Identify candidates for optimization or review | | Investigate spikes in failures or terminations | Prevent cascading failures in automated workflows | | Monitor most frequent tasks | See what’s driving system load and how to prioritize scaling | | Filter by project/team | Hold teams accountable for clean and efficient automations | | Review success rate over time | Track improvement or degradation in platform reliability | *** ### Example Admin Use Case Suppose a team reports that “document ingestion is slow this week.” 1. Open the Task Monitor tab. 2. Filter by that team’s project ID. 3. Look at “Top Time-Consuming Tasks” to find slow workflows. 4. Use “Task Duration Range” to identify spikes. 5. If failures are involved, go to “Task Execution Distribution” and “Overall Success Rate.” You’ve now isolated the issue, confirmed whether it’s task-related, and can decide whether to contact the team, escalate to DevOps, or adjust configurations.