Configure an agent for the Slack app (Beta)

This documentation describes one or more private beta features that are in development. Beta features are subject to quick, iterative changes; therefore the current user experience in the Sigma service can differ from the information provided in this page.

This page should not be considered official published documentation until Sigma removes this notice and the beta flag on the corresponding feature(s) in the Sigma service. For the full beta feature disclaimer, see Beta features.

If you are interested in joining a limited test group and enabling this feature in your Sigma organization, contact Support or reach out to your Account Executive.

The use of AI features is subject to the following disclaimer.

You can configure Sigma Assistant, a Snowflake Cortex Agent, or a Databricks Genie Space to perform analysis on your data from Slack. Once configured, the agent can answer questions about your data, share visualizations, and start analysis in Sigma from within a Slack conversation.

System and user requirements

To configure a Slack agent, the following requirements apply:

You must be assigned the Admin account type.
You must configure a Slack integration for your organization.
To use Sigma Assistant as a Slack agent, you must set up an AI provider for your organization.
To use a Snowflake Cortex Agent or Databricks Genie Space as a Slack agent, you must configure a warehouse agent that is available to all users in your organization.

To use Sigma Assistant in Slack, the following requirements apply:

You must be assigned an account type with the Use Assistant permission enabled. See Account type and license overview for details.

Notes

Sigma Assistant only answers questions in Slack based on configured data sources.
In a thread, the agent only responds when directly mentioned by members who have connected their Sigma account to Slack. If you reply in a thread and the agent doesn’t respond, connect your account and try again. See Connect your Sigma account to Slack.

Configure a Slack agent for an organization

To configure the Slack agent:

Go to Home > Administration > AI settings.
Select the Assistant tab.
In the Assistant integration section, next to Slack, select Add.
For Agent, select one of the following:
- Sigma Assistant — Sigma finds a data source in your catalog to answer each question.
- A Snowflake Cortex Agent or Databricks Genie Space configured in your organization.
When using a warehouse agent, such as a Snowflake Cortex Agent or Databricks Genie Space, in Slack, the selected agent must be available to all users.
If the agent you want to use for the Slack integration does not appear in the list of available agents, you must grant access to all users in your organization. For more information, see Grant access to a warehouse agent in Sigma.
Select Save.

The agent is available to your Slack integration.

Connect your Sigma account to Slack

After configuring a Slack agent for your organization, you can authenticate to connect your Sigma account to Slack. Once authenticated, you can ask questions and query data based on the permissions associated with your Sigma account.

To authenticate to your Sigma account from Slack:

In Slack, send a message to the Sigma Notifications Slack app.

The Slack app responds with a prompt to connect your Sigma account.
Select Connect Sigma.

The Connect Slack page opens.
After reviewing the details on the Connect Slack page, select Allow.

Your Sigma account is connected to Slack.

Use the Slack agent

You can use the Slack agent to ask natural language questions about your data, share visualizations, and start analysis in Sigma.

Ask about your data

In a chat with the Sigma Notifications Slack app, you can send a message with a question about your data to get analysis back from the agent.

For example, if your agent has data on short-term rental prices, you can ask a question like the following:

What's the average price per night for rentals by city?

The agent responds with a list of average prices per night, grouped by city.

Ask about a specific data source

By default, the agent searches your catalog for a data source that can answer your question. To scope a question to a specific source, include a link to that source in your message.

Paste the URL into your Slack message along with your question. For example:

https://app.sigmacomputing.com/<org-name>/data-model/<data-model-url-id>
What were our top five regions last quarter?

You can link to a table, a data model, a Snowflake semantic view, or a workbook. If the link points to another type of object, or to a source you don’t have access to, the agent searches your catalog instead.

Create visualizations

In a chat with the Sigma Notifications Slack app, you can send a message asking the agent to create a visualization and share it in the Slack thread.

For example, if your agent has data on electronics sales, you can ask a question like the following:

Can you send me a bar chart showing sales of cameras in the East over the past year?

The agent creates a workbook with the chart and responds with a link to the workbook along with an exported image of the chart.

Continue analysis in Sigma

In a chat with the Sigma Notifications Slack app, you can send a message asking the agent to start analysis, then follow the provided link to a new exploration in Sigma.

For example, if your agent has data on flights, you can send a prompt like the following:

I want to explore a table showing average flight time over the last 6 months.

The agent creates a new exploration with the data and responds with a link to view it in Sigma.

Continue a conversation in a thread

In a direct message with the Sigma Notifications Slack app, the agent replies in a thread to each message you send.

In a channel or group conversation, @-mention the app to start a conversation. The agent replies in a thread. In a thread, the agent responds only when you directly mention it, so @-mention the app again to continue the conversation.

View the steps behind an answer

As the agent works on your question, it posts each step it takes. When the answer arrives, select Show plan to open the Assistant plan dialog, which lists the steps the agent took and the queries it ran.