--- title: "Pigment AI" slug: "pigment-ai" description: "Explore Pigment AI's tools like the Navigation, Data, and Documentation Assistants, Insights, and more." tags: ["applications", "boards", "Data Assistant", "Forecasting", "insights", "machine learning", "Navigation Assistant", "pigment ai"] updated: 2026-08-06T08:02:19Z published: 2026-08-06T08:02:19Z canonical: "kb.pigment.com/pigment-ai" --- > ## Documentation Index > Fetch the complete documentation index at: https://kb.pigment.com/llms.txt > Use this file to discover all available pages before exploring further. # Pigment AI Pigment AI is a suite of state-of-the-art AI capabilities designed to help you build, forecast, analyze, and visualize your business data. At its heart are AI Agents — your specialized digital workforce. They understand natural language prompts in more than 50 languages and include the following specialized agents: The **Modeler Agent** helps you create new Applications, Blocks and Boards, and turn data into visuals with Pigment Frames. The **Analyst Agent** is more powerful than ever, chatting with Members, finding Reports and Boards, pulling trends, insights and conclusions from your data and helping you design repeatable, schedulable Missions. The **Custom Agent** lets you combine bespoke company knowledge with Pigment's powerful AI capabilities, navigating Boards and analyzing data more effectively. And the **Documentation Agent** accepts user help queries about the platform and issues short summaries and links to Pigment’s Knowledge Base. Other features powered by Pigment AI remain unchanged: - **Predictions** generates customized forecasts automatically for inclusion in your model. See [Machine Learning Forecasting](/v1/docs/machine-learning-forecasting) for more information. - The **MCP Server** allows external AI systems to connect to Pigment’s features and data sources in a controlled, secure, and structured way. See [MCP Server](/v1/docs/mcp-server) for more information. - **Explain the formula**, found by the formula bar, opens the Documentation Agent with the current formula pre-loaded. > [!TIP] > 🎓 **Pigment Academy** > > - For detailed insights, examples and use cases on Pigment’s AI features, explore the [Pigment AI Learning Path](https://academy.pigment.com/page/pigment-ai-catalog). > - To learn more about the Agentic Workforce, check out the [Introduction to Pigment’s Agentic Workforce](https://academy.pigment.com/introduction-to-pigments-agentic-workforce/532266) video. > - For a video in English or French on best practices when setting up for AI, see [Getting your models ready for Pigment AI](https://academy.pigment.com/path/pigment-ai/getting-your-models-ready-for-pigment-ai/437096). Just make your language selection in the player! ## Set up Pigment AI Follow these steps to set Pigment AI up for your Application. 1. **Request activation** The **Workspace Administrator**, **Security Administrator** or the **Primary Owner** must submit a Support ticket [here](https://support.pigment.com/en/support/tickets/new?ticket_form=ai_feature_request) requesting activation. In your ticket, under **AI Feature Request**, specify your choice of the below: > [!NOTE] > ℹ️ **Note** > > MCP Server activation is included when you choose any of the above. - “Modeler Agent, + Analyst, Custom & Documentation Agents” - “Analyst Agent, + Custom & Documentation Agents” - “Documentation Agent” - “MCP Server” - “Predictions” using Machine Learning Forecasting 2. **Grant Members access using dedicated permissions** Two permissions allow you to manage which Members have access to the Agents: - `Access AI Agents` - grants access to: - Modeler Agent (also requires `Configure Blocks`) - Analyst Agent - Custom Agent - Documentation Agent - `Configure AI Agents` - grants permission to: - persist and automate analyses via Missions - create and configure Custom Agents By default, all Members with `Define Application Security` are automatically granted `Configure AI Agents`. For more information on updating Permissions, see [About Roles, Permissions, and Access Rights](/v1/docs/roles-permissions-access-rights#permissions). The table below sets out the permissions needed by AI features: | Feature | Permissions needed to configure | Permissions needed to use | | --- | --- | --- | | **Modeler Agent** | No configuration required | `Configure Blocks` and `Access AI Agents` | | **Analyst Agent** | `Configure AI Agents` | `Access AI Agents` | | **Custom Agent** | `Configure AI Agents` | `Access AI Agents` | | **Documentation Agent** | No configuration required | `Access AI Agents` | | **Predictions** | `Define Application security` | | **MCP Server** | Requires one of these Roles: - Security Administrator - Workspace Administrator - Primary Owner For more information, see [MCP Server](/v1/docs/mcp-server#how-to-connect-to-the-pigment-mcp-server) | None | This means that: - Anyone with `Access AI Agents` can use any AI Agent that your Application supports, except for the Modeler Agent. - Anyone with `Configure Blocks` and `Access AI Agents` can use the Modeler Agent. To take Modeler Agent access away from a Member, you need to remove their `Configure Blocks` permission. - If at Workspace-level the Modeler and Analyst Agents are activated, you can not give access to the Modeler Agent without also giving access to the Analyst Agent. As a rule, the Agents inherit permissions and data access rights from the user interacting with them, ensuring that they can only view data and perform actions the user is authorized for. #### Update Roles as needed Follow these steps to update Roles needing access: 1. In your Application, select the three dots at the bottom of the sidebar and then **Roles, permissions & access**. 2. The main pane opens on Overview. Select the next tab over: **Roles**. 3. Hover the mouse over the Roles needing access. Select the blue pencil icon that pops up to **Edit Role**. In the dialog that appears, under **Permissions**, open the drop-downs as appropriate, toggling **On** the permissions according to the grid below. 4. **Save**. > [!NOTE] > ℹ️ **Note** > > By default, all Roles have `Access AI Agents` and Admin Roles have `Configure AI Agents`. For more information, see [About Roles, Permissions, and Access Rights](/v1/docs/roles-permissions-access-rights#permissions). ## AI data access The **AI data access** setting allows you to prevent the MCP Server, the Analyst Agent, and the Custom Agent from querying data in a particular Block. This setting does not affect the Modeler Agent. Once the MCP Server, the Analyst Agent, and the Custom Agent are activated, they have access to all an Application’s Blocks by default. You may want to hide certain Blocks from the MCP Server, the Analyst Agent, and the Custom Agent, for example if they contain sensitive data. ### Change the setting You need two permissions to operate this setting: `Configure AI Agents` and `Configure Blocks`. You can adjust it for individual or multiple Metrics. **Individual Metrics** To change the setting for just one Metric, open its settings and scroll down in the **General** tab to **AI data access**. For more information, see [Block settings for Metrics](/v1/docs/block-settings-for-metrics#ai-data-access). **Bulk update Metrics** To update this setting for multiple Metrics: 1. In your Application, open the **Blocks** folder in the left sidebar and select **All Blocks**. 2. Select the **Metrics** tab. 3. (*Optional*) In the row below, open the **AI** drop-down menu on the right. Select **Filter** if it is not displayed. Choose from the following options: 1. Can access Metric’s data 2. Cannot access Metric’s data 4. Select the checkbox next to **Name** to select all Metrics, or use the individual checkboxes to select specific Metrics. The bulk action toolbar appears. 5. Select **Toggle AI data access** from the bulk action toolbar. 6. Select **Allow AI access to Metric’s data** to confirm. ### Types of actions affected Turning off access strictly impacts Block data querying, and does not prevent the following types of actions: - Block structure modifications - Block configuration edits - Adding or editing Views - Any non-data actions > [!NOTE] > ℹ️ **Note** > > The Analyst Agent and Custom Agent can access Metrics, Transaction Lists, and Dimension Lists. For Dimension Lists: > > - When **AI data access is toggled on**, the Agents can retrieve all Dimension List Property values when querying the List directly. > - When **AI data access is toggled off**, Property values are not accessible. > > However: > > - If a Metric or Transaction List references a Dimension List, the Agents can still see the **Item names** from that List. > - No additional details are exposed. > - All access is still subject to access rights. The Agents inherit data access rights from the user interacting with them, ensuring that they can only see Item names and other data the user is authorized for.