# Agent Builder

To start creating your first Agent with Agent Builder, simply log in to nexos.ai. Then, for quick access, navigate to the plus sign next to Agents or click on the Agent icon in the navigation bar and click on the “Create agent from scratch” button in the drop-down menu. Here are the best practices for using Agent Builder to create the best AI teammate. 

{% stepper %}
{% step %}

### Describe the task you want to delegate

Tell the AI Agent Builder what you need in plain language. Just describe the job as you would to a new human hire. Helpful guiding questions for your first prompt: 

* What should your Agent do? 
* What decisions should it make?
* What should it avoid? 

The clearer your description, the better your Agent performs. Think about the actual problem you want to solve. For example, “I want to stay on top of things I need to do – please review my emails, Jira, and Slack threads, and compile a list of action items every morning at 8 AM.” If you need the Agent to run on a particular schedule (for example, send you an overview of all Slack updates every morning), ask it to run periodically. Press “Send.”
{% endstep %}

{% step %}

### Refine your agent

The AI Agent Builder asks follow-up questions and sharpens the instructions until your Agent's behavior is exactly right. You review, approve, and move on. This back-and-forth ensures your Agent matches your exact workflow. You stay in control. If something isn't quite right, you can iterate until it's perfect. 

At this stage, the Builder will propose Agent name, description, enable or disable Web Search, draft starters, and show Flow to your right – a visual representation of the steps your Agent will take to complete the task at hand, with optional steps for human-in-the-loop reviews. Look over the process closely and suggest edits if necessary.

Some parameters, like creativity level, are automatically selected using best practices, but all settings can be changed if needed via prompt or Agent editor.
{% endstep %}

{% step %}

### Connect your tools

Type the integrations your Agent needs: Slack, Jira, Google Workspace, SharePoint, and more. Your Agent works within the tools you already use, pulls information from the systems your team relies on daily, and acts within this context. 

The Agent Builder will suggest integrations that make sense for your Agent. This step is optional – if your Agent doesn’t need access to tools or internal data, skip it.
{% endstep %}

{% step %}

### Test it

Try prompting your Agent straight away in the “Try Agent” window on the right to see it in action and make any quick adjustments. Run it on real work scenarios. 

Here are a few guiding questions for testing and final refinement: 

* Does it handle edge cases? 
* Is the tone right? 
* Does it pull the right data? 

Testing is fast. Adjustments are faster. You'll know immediately if your Agent is ready.
{% endstep %}

{% step %}

### Launch your agent

Your Agent is ready. Click “Save” – now, it’s available for you anytime to handle the work you've been doing manually. The page will redirect immediately to the chat with your Agent or, if it’s a scheduled Agent, to the list of all Agents.
{% endstep %}

{% step %}

### Edit your agent

If you need to further edit your Agent at a later time, navigate to Agents > Created by you > Three dots on the right > Edit. Here, you can customize every part of your Agent, like changing the model, schedule, starter prompt, name, description, and more.
{% endstep %}
{% endstepper %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.nexos.ai/workspace/agents/agent-builder.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.