You start by giving the agent a simple instruction: 💬 Check tasks in Jira marked "In Review". If everything is done, close them.
And then the magic starts:
1) The agent’s categorizer first analyzes the prompt to understand the intent: it's a Jira workflow task involving review and resolution. Then, it identifies the key subtasks.
2) Next, the agent’s planner breaks subtasks into actions: logging into Jira, navigating to the project, finding tickets and interpreting content, testing task, reviewing code and finally resolving task.
3) The executor agent then steps in, performing each of these actions in your browser — just like a human would, but faster.
What looks like a single sentence is actually an orchestrated sequence of decisions and actions. The agent reads it, understands it, and gets it done.
Agent actions step by step
1. Planning workflow
The agent analyzes users prompt and breaks it into subtasks, then transforms every subtask into action.
2. Applying styles
The AI agent formats your document with heading styles, tables, and a clear structure.
3. Cleaning data
The agent removes duplicates, organizes sheets, and prepares error-free datasets.
4. Generating summary
The agent reads your data, extracts key insights, and writes a concise TL;DR.
5. Exporting results
The agent adds the TV to the cart, confirms address and payment details, and completes the checkout.
6. Reporting back
After finishing the task, agent writes back to user in chat window and confirms that the task is completed.
Be the first to try Sigma
We’re rolling out early access now. Join the waitlist and get notified as soon as your spot is ready.