> ## Documentation Index
> Fetch the complete documentation index at: https://docs.runspark.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Agent Settings

> Configure your agent's name, title, avatar, and AI model

Your agent's settings control its identity and the AI engine behind it. You can access these from the **Agent Settings** tab on your agent's detail page.

***

## Name & Title

### Name

This is what your agent calls itself — and what you'll see in the sidebar, chat, and notifications. Pick whatever fits your business:

* **Professional:** "Alex," "Morgan," "Riley"
* **Fun:** "Ziggy," "Nova," "Spark"
* **Descriptive:** "Sales Assistant," "Support Bot"

Your agent's name appears everywhere it shows up — in chat, on the dashboard, in notifications, and if connected to messaging apps like Telegram or Slack.

### Title

A short description that appears below the name. This helps you (and your team) quickly understand what this agent does, especially if you have multiple agents:

* "Event Coordinator"
* "Insurance Sales Assistant"
* "Lead Qualification Specialist"
* "Marketing Content Creator"
* "Customer Support Agent"

> **You:** Change your name to Sam and your title to "Senior Account Manager."
>
> **Agent:** Done! I'm now Sam, Senior Account Manager. Updated my display name and title.

***

## Avatar

Upload a profile picture for your agent. This appears in the sidebar, the agent detail page, and in the chat panel. It's a small thing, but it makes your agent feel more like a team member.

* **Supported formats:** PNG, JPEG, WebP
* **Maximum file size:** 2MB
* **Tip:** A clean, simple image works best since it's shown at small sizes

You can upload an avatar from the agent settings page by clicking on the avatar area and selecting an image.

***

## AI Model

Your agent uses an **AI model** to understand your messages, reason through problems, and generate responses. Think of the AI model as your agent's "brain" — different brains have different strengths.

### What Is an AI Model?

An AI model is the technology that powers your agent's intelligence. Different companies build different models, and each one has trade-offs:

| Provider      | Model Family | Best For                                                                                               |
| ------------- | ------------ | ------------------------------------------------------------------------------------------------------ |
| **Anthropic** | Claude       | Following complex instructions, maintaining consistent personality, nuanced writing, careful reasoning |
| **OpenAI**    | GPT          | Versatile general use, coding tasks, creative writing                                                  |
| **Google**    | Gemini       | Fast responses, competitive pricing, broad knowledge                                                   |

### Which Model Should I Choose?

If you're not sure, **start with the default recommendation** — it's been selected because it works well for most business use cases. You can always change it later.

Things that might lead you to switch:

* **Your agent isn't following instructions well** → Try Claude (Anthropic). It's particularly good at following complex, multi-step rules consistently.
* **Responses are too slow** → Try a faster model tier. Some models have "fast" variants that sacrifice a bit of depth for speed.
* **You're budget-conscious** → Different models have different costs per message. Gemini tends to be more affordable.

### API Keys

To use an AI model, your workspace needs an API key from the model provider. This is a code that gives your agent permission to use the provider's service. You can add your API key during workspace setup or in **Settings → API Keys**.

<Info>
  **If you're not sure what an API key is:** It's like a subscription credential. When you sign up with an AI provider (like Anthropic, OpenAI, or Google), they give you a key — a long string of letters and numbers — that you paste into Spark. This connects your agent to their service. Each provider has their own pricing based on usage.
</Info>

***

## Restarting Your Agent

Sometimes you need to restart your agent — for example, after making significant changes to its personality, instructions, or knowledge files. Restarting refreshes the agent's runtime.

**What restarting does:**

* Refreshes the agent's session
* Re-reads all configuration files (personality, instructions, knowledge)
* Clears the current conversation context (but NOT memory — your agent's long-term memory and daily notes are permanent)

**What restarting does NOT do:**

* Does not delete any data
* Does not erase memory or daily notes
* Does not remove knowledge files
* Does not delete schedules or automations

**When to restart:**

* Your agent seems "stuck" or isn't responding properly
* You've made major changes to personality or instructions and want them to take effect immediately
* You've updated knowledge files and want a fresh start
* Something just seems "off" and you want to reset

You can restart from the agent detail page by clicking the **Restart** button, or just ask:

> **You:** Restart yourself.
>
> **Agent:** Restarting now. I'll be back in a moment with a fresh session.

<Warning>
  Restarting clears the current conversation context — meaning your agent won't remember what you were *just* talking about. But everything in its memory files (daily notes, long-term memory) is preserved. It's like your agent stepping out of the room and coming back in — same person, same knowledge, just a fresh start to the conversation.
</Warning>