> For the complete documentation index, see [llms.txt](https://docs.soda.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.soda.io/soda-documentation/soda-ai/soda-ai-agent.md).
# Soda AI Agent
Ask questions, get guidance, and let Soda perform actions for you in natural language, without leaving Soda Cloud.
The Soda AI Agent is **the in-product, conversational face of Soda AI**. It lives alongside your work in Soda Cloud and helps **you understand data**, **author data contracts**, and **accelerate onboarding** without replacing human decision-making.
You can request it to **perform actions** on your behalf, get **step-by-step guidance** on platform workflows, or ask it **questions** about Soda, all in **natural language, without leaving the page you're on**. As an agentic assistant within Soda Cloud, **the Soda AI Agent can access the tooling** necessary to perform actions on your behalf, such as Contract Autopilot or Soda MCP.
## Chat window
The Soda AI Agent sits in a chat window that stays at the bottom of every page in Soda Cloud. Type your request in the **Ask Soda AI anything...** field and send it.
### Capabilities
The Soda AI Agent's actions are scoped to the **same operations you can perform** yourself in Soda Cloud, subject to your role and permissions. **The Soda AI Agent can:**
* **Perform** safe, reviewable **actions in Soda Cloud** on your behalf. For example:
* **Generate data contracts.** Using Contract Autopilot, you can point the Soda AI Agent at your datasets and generate fully populated contracts with recommended checks, drawn from your data's own profile. Data teams can get full coverage in an afternoon, instead of manually creating contracts for weeks.
* **Edit and iterate over data contracts** in plain English. Add a freshness check, tighten a rule, handle a nested field, all without hand-writing code. Within the chat, you will see exactly what changed, so you keep full agency over the editing process.
* **Guide** **you through platform workflows** (configuring datasets, authoring contracts, investigating incidents)
* **Answer product and docs questions.** Ask what a check, monitor, or incident means, or tell it to make a change. It proposes the action, scoped to what your role already allows, and waits for you to approve.
* **Agentic navigation.** Ask the Soda AI Agent to take you to another place in the Web UI and it will take you there.
* **Context awareness.** The Soda AI Agent knows basic information about the page you are visiting to be better able to answer related questions.
{% hint style="info" %}
Every change is proposed, shown, and approved by a human user. The Soda AI Agent is an assistant and control surface, and is designed to perform **human-in-the-loop processes**.
{% endhint %}
### Conversations and history
Each chat is saved as a **conversation** in the Soda AI Agent chat window. You can resume a previous conversation or start a new one from any page on Soda Cloud.
{% hint style="info" %}
**Soda does not store prompts or conversations.** Your prompts and conversation history are stored in your browser local storage. If you change browsers or open Soda Cloud in a different device, you will not have access to your conversation history.
{% endhint %}
***
## Data usage and privacy
* To answer questions and propose or perform actions, the Soda AI Agent analyzes your **natural-language request**, Soda **documentation content**, basic information about the **page you are on**, and, when needed, **dataset schemas** and **non-identifiable column-level metadata**.
* The Soda AI Agent works within an **authenticated Soda Cloud session** only.
* Every action taken by the Soda AI Agent is subject to your **existing Soda roles and permissions**.
* When the Soda AI Agent uses [Contract Autopilot](/soda-documentation/soda-ai/contract-autopilot.md) or [Contract Copilot](/soda-documentation/soda-ai/contract-copilot.md) to carry out a request, those features send the data described on their own pages.
* If your organization brings its own key (BYOK), Soda AI Agent requests go to your own OpenAI project instead of Soda's.
***
## Enable/disable the Soda AI Agent
An organization admin enables or disables the Soda AI Agent for the whole organization.
{% stepper %}
{% step %}
Navigate to **Settings** > **Organization Settings**.
{% endstep %}
{% step %}
Under the **General** tab **> AI** tab, check or uncheck **Enable Soda AI Agent**.
{% endstep %}
{% step %}
Click **Save** on the top right corner.
{% endstep %}
{% endstepper %}
***
{% hint style="info" %}
You are **not logged in to Soda** and are viewing the default public documentation. Learn more about [Licensing & documentation access](/soda-documentation/govern/security-licensing-and-compliance/documentation-access-and-licensing.md).
{% endhint %}
---
# 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.soda.io/soda-documentation/soda-ai/soda-ai-agent.md?ask=&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.
