# How SpringBrand works

Connect your Agent, describe a goal, inspect the selected capability, and use the result to guide the next run.

SpringBrand connects your existing Agent to professional Plugins and APIs. Tell your Agent what you want to achieve. SpringBrand selects fitting capabilities, runs the task, reviews the result, and uses that feedback to improve later choices. The exact operations available depend on your account and Agent connection.

## The request-to-result flow

1. Connect your Agent

   Install the SpringBrand integration for your client, complete SpringBrand sign-in, and open a fresh session.

2. Describe the outcome

   State the task, audience, inputs, constraints, and what a useful result would look like.

3. Discover and select a capability

   Ask your Agent to inspect the available SpringBrand operations. Check why the chosen Plugin or API fits, its required inputs, and any account access.

4. Run with the right approval

   Provide missing information and approve requested access or external action when appropriate. If no matching operation is exposed, the Agent should say so.

5. Review and continue

   Inspect sources, dates, limitations, and completed actions. Use the result to decide the next task or refine your request.

## Intelligent selection and a learning loop

SpringBrand matches your task to fitting APIs and Plugins, then reviews the result to refine later choices. For a specific run, ask your Agent which operation it selected and compare the output with your goal.

## Keep tasks moving beyond a chat session

SpringBrand can keep supported tasks running in the cloud, 24/7, even when your Agent is offline. Before leaving a task unattended, check how to start, monitor, and stop it in your account, and review any approval required for external actions.

> Your Agent’s live tool list shows the operations you can use and their required inputs and access. Check it when you choose a Plugin or start a background task.

- [Getting started](/developers/docs/quickstart): Connect and verify an Agent.
- [Smart tool matching](/developers/docs/tool-routing): Inspect the capability chosen for a specific task.
- [Keep working 24/7](/developers/docs/continuous-execution): Check available controls before relying on unattended work.