> For the complete documentation index, see [llms.txt](https://docs.firefly.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.firefly.ai/key-features/ai-assistant.md).

# AI Assistant

Firefly incorporates an AI Assistant, affectionately called your "Thinkerbell AI," which brings conversational intelligence to cloud management and automation. This feature allows you to interact with your cloud infrastructure using natural language queries and remediate issues with a single click.

Thinkerbell Chat is a working surface, not just a chat box. You can see every step the agent takes, come back to any past conversation, and start a long job and close the window while the agent finishes it in the background.

## Opening Thinkerbell

* **From any page:** Click **Ask Thinkerbell anything...** in the app header, or press **⌘ J** on Mac or **Ctrl J** on Windows and Linux. Thinkerbell opens as a modal. You can dock it to a side panel on the right to keep it open while you work, and expand it back to a full modal.
* **AI SRE:** Search for **AI SRE** in the Firefly search to open Thinkerbell as a full-page chat.
* **Self-Service:** In **Workflows > Self-Service**, the **Generate with Thinkerbell AI** tab uses the chat to generate IaC. See [Self-Service Infrastructure](https://docs.firefly.ai/key-features/infrastructure-as-code-automation.md#self-service-infrastructure).

Thinkerbell follows the app theme, including light mode, and switches as soon as you change the theme.

## Natural Language Queries

The AI Assistant allows you to ask questions in plain English about your cloud and get instant answers. This is incredibly useful for quick insights or troubleshooting without digging through consoles and dashboards. The AI is context-aware, it knows about your cloud inventory, configurations, and even recent events.\
Check out the [Firefly MCP server](https://docs.firefly.ai/integrations/mcp.md) for more details on how to use the AI Assistant in your IDE.

### Example Queries

You can ask questions like:

* "How many EC2 instances do we have running in region us-west-2?"
* "Which IAM users do not have MFA enabled?"
* "Do we have any unmanaged S3 buckets with public access?"

The assistant will parse your question, run the necessary searches or analysis in Firefly, and respond with a concise answer.

### Suggested Prompts

A new chat opens on a welcome screen with up to three suggested prompts. Click a prompt to send it. The prompts match the page you're on. For example, on the Inventory page you can ask Thinkerbell to identify drifted unmanaged resources, and on a workspace you can ask it to explain recent run failures.

### Page Context

The chat box shows the page or resource you're on as a context chip, so Thinkerbell knows what you're looking at. To ask a question without it, remove the chip with its **Remove page context** button. To bring it back, click **Add page context**.

The chat box grows as you type. Press **Enter** to send, or **Shift + Enter** to start a new line.

## Reading the Answers

Answers read like a clean document, in a centred column with the Firefly mark beside them:

* **Code** appears in cards that show the language, such as Terraform, HCL or YAML. Click **Copy** to copy the code.
* **Tables** show asset states as status pills: **Codified**, **Unmanaged**, **Drifted**, **Ghost** and **Discovered**. See [Terminology](https://docs.firefly.ai/introduction/terminology.md) for what each state means.

### Agent Steps

Thinkerbell shows what the agent is doing while it works. While a tool runs, the step shows its progress, for example **Running Inventory…**. When the agent finishes, the step shows a summary, for example **Ran 3 tools · 4.2s**. If a tool fails, the summary shows how many failed.

Click the summary to expand it and see each tool call, with its arguments and status.

## Background Runs

The agent keeps working after you close the chat window. Start a long job, close Thinkerbell, and come back later. When you reopen the chat, it's in the history sidebar. If the agent prepared code changes, the chat shows the **Create Pull Request** button, and after you open the pull request, a **View Pull Request** link.

## Chat History

Every Thinkerbell surface has a chat history sidebar:

* Chats are grouped by **Today**, **Yesterday**, **Last 7 days** and **Older**.
* Click **New chat** to start a new conversation, or **Search chats** to find a past one by its title.
* A dot marks chats that are still running and chats that finished while you were away.
* Click a past chat to open it. It becomes the current chat again, so you can pick up where you left off.
* To delete a chat, hover over it, click the delete icon, and confirm. A deleted chat stays deleted.
* Click **Collapse sidebar** to make more room for the chat, and **Expand sidebar** to open it again. In a narrow panel, the sidebar collapses to a slim rail.

Each Firefly account has its own chat history. When you switch accounts, you see only that account's chats and agent activity.

## Remediation

The AI Assistant can also suggest remediation steps for different issues detected by Firefly, such as Governance policy violations, Workflows Guardrails violations, etc.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

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

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.firefly.ai/key-features/ai-assistant.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.
