Creating Tasks

14 January 2026•4 min read

Tasks are reusable workflows that define what your agent should do and when. Instead of giving the same instructions repeatedly, you can create a task once and run it whenever needed—manually, on a schedule, or at regular intervals. Tasks transform one-off requests into repeatable automation.

The power of tasks lies in their flexibility. You can describe what you want in plain language, and your agent figures out the technical details of how to accomplish it. You don't need to specify every step or understand the underlying mechanics. Just explain the goal, and your agent takes it from there.

Creating Tasks from Natural Language

The easiest way to create a task is simply to tell your agent what you want done, including when it should happen. Your agent understands scheduling language and will set up the task appropriately.

Examples of natural language task creation:

"Every Monday at 9am, check my LinkedIn messages and summarize new connection requests"
"Research competitors weekly and compile a report"
"Post to Twitter every day at noon with industry news"
"At the end of each month, pull our sales data and create a summary spreadsheet"

When you give instructions like these, your agent recognizes that you want a recurring task. It creates the task, sets up the schedule, and begins executing at the specified times—all from a single conversational request.

Agent-Created Tasks

Your agent isn't limited to following explicit task creation requests. It can recognize patterns in your work and proactively suggest turning recurring needs into scheduled tasks.

If you frequently ask for similar work—checking the same website, generating the same type of report, performing the same research—your agent might suggest creating a task to automate it. This proactive behavior helps you capture automation opportunities you might not have thought to formalize.

The agent can also create tasks during complex workflows. If it identifies something that should run regularly as part of a larger goal, it might set up that recurring piece automatically while working on the immediate request.

Manual Task Creation

You can also create tasks directly through the interface without going through conversation. This is useful when you want more control over the configuration or when you're setting up multiple tasks at once.

To create a task manually:

Open your agent's task list
Click "Create Task" or "Add Task"
Give the task a descriptive name
Write a description of what the task should accomplish
Set when it should run (manual trigger, scheduled time, or recurring interval)
Save

The description you provide becomes the instructions your agent follows. Write it as you would explain the task to a capable assistant—clear about the goal, specific about any constraints, but not prescriptive about the exact steps.

Task Templates

For common workflows, O-mega provides pre-built task templates. Templates give you a starting point with proven instructions that you can customize for your specific needs.

When you add a task from a template:

Base instructions come pre-configured from the template
You can add your own custom instructions on top
Templates receive updates over time while preserving your customizations

Templates exist for common use cases like social media monitoring, competitive research, data collection, and report generation. They encode best practices discovered across many users.

How Tasks Execute

When a task runs—whether triggered manually, by schedule, or by interval—your agent approaches it like any other request. The difference is that the instructions come from the task definition rather than a live chat message.

The execution flow:

Your agent receives the task description and any additional context
It plans the necessary steps to accomplish the goal
It executes using whatever capabilities are needed—browser sessions, computer sessions, or connected tools
Results are delivered and any outputs are saved as deliverables

The agent works autonomously, deciding the best approach based on your description. A task that says "research competitors and create a report" doesn't need to specify whether to use browser sessions or computer sessions—your agent figures that out.

This autonomous intelligence means your tasks remain effective even as the underlying websites or data sources change. Your agent adapts its approach to accomplish the goal rather than blindly following predetermined steps.