Sharing information between departments or business units ("domains") of a company is critical for success. Sharing and consuming data and AI assets is more successful if they are shared as a product. A data product is a collection of related data objects plus metadata, such as a business description, ownership and contact information, service level objectives, data dictionary, and more. In a Data Mesh, data products are typically also subject to various data management and organizational principles.

Snowflake Internal Marketplace enables companies to publish documented and governed data products, to make them discoverable and understandable for data consumers. A previous hands-on lab is available to exercise the internal marketplace and learn how governance policies can control access to data products. In this lab we briefly recap the basics and then focus on sharing AI assets such as agents, semantic views, and other new features.

Data Products consist of:

Your Snowflake account has been pre-configured with a database, users roles, and organizational profiles you need for this lab. In this section you will review the environment before creating your first data product in the next section.

The lab environment consists of:

Log in to your account as the provider user.

1. Review the database: In the menu on the left, click on Catalog > Database Explorer. Open the TPCH database and navigate to the SF1 schema. You should see the TPC-H tables, the ORDER_SUMMARY view, and the ORDERS_PER_CUSTOMER function.

2. Review the Internal Marketplace: In the menu on the left, click on Catalog > Internal Marketplace. You will see 1 or more listings (data products) that we have pre-created. Open any one of them to examine their structure content.

3. Review the organizational profiles: While viewing the Internal Marketplace, use the Profile filter to see the available domain profiles. (At the start of the lab, filtering for a specific profile will likely return 0 data products.)

Open a new file in a workspace (or a worksheet) and run the following commands to confirm the lab setup:

You are now ready to begin the lab exercises!

In this section you will create and publish your first organizational listing called Order Insights. This data product shares order and customer data from the TPC-H database with other business domains in your organization.

You will work as provider with the provider_role. The publishing flow in the Provider Studio consists of 5 steps.

1. Log in to your account as the provider user.
2. Navigate to Catalog > Internal Marketplace.
3. Click the blue + Create Listing button in the top right.
4. Proceed with "AI generation on" or, optionally, turn it off. "AI generation on" will automatically populate some of the data product metadata for you.
5. Click +Select to select the data objects to share

Now select the data objects for this data product.

1. Navigate to the TPCH database > SF1 schema, and select:
   • These 5 tables:
     • CUSTOMER
     • LINEITEM
     • ORDERS
     • REGION
     • SUPPLIER
   • The ORDER_SUMMARY secure view
   • The ORDERS_PER_CUSTOMER function
2. Click Done to proceed, then Generate Listing if you chose "AI generation on"

1. Review the AI-generated listing metadata
2. Change the listing title to "Order Insights (<account locator>)"
   • For disambiguation in this lab only, add your 7-character Snowflake account locator to the listing title. The locator is part of the URL in your browser address bar.
3. Under title use the Select profile box to select the profile (domain) under which to publish this data product
   • You're free to pick any profile from the list - there is no right or wrong in this lab.
   • Among all the lab attendees we hope to see many different profiles being chosen!
4. If desired, make changes to the listing description or other metadata fields.
   • For example, in the Documentation field you could add any URL to supplementary documentation, e.g. http://www.snowflake.com.
5. Enable Data preview in the Data dictionary.
   • Data preview is off by default for privacy reasons.
   • Let's enable data preview as shown below.

6. Sensitive column masking in the data preview.
   • Click the Edit button at the top right of the data dictionary.
   • Select the customer table.
   • Select the c_address, c_name, and `c_phone columns for masking. Save.

7. Set custom attributes to any values you like:
   • Click in the field "Attributes (optional)".
   • Note that the attributes Confidentiality and Source Systems are examples of Custom Attributes that the internal marketplace admin can define for your company.
   • The attribute Source Systems is multi-valued, i.e. you can assign mutiple values from the drop-down list.
   • Custom attributes can also appear in the markeplace UI as filters.
   • Update frequency is a standard attribute that you can set to indicate the intended service level objectve in terms of data product freshness.

Click the ">" button next to "Discovery" to configure who can discover and access this listing.

Configure this listing as follows:

Click Set up request approval flow to proceed.

Click the blue Publish button in the top right corner.

Your Order Insights data product is now live on the Internal Marketplace!

To verify, navigate to the Catalog > Internal Marketplace.

In this section:

1. Log in to your account as the consumer user (e.g. in a new incognito browser window).
2. Navigate to Catalog > Internal Marketplace.
3. You should see your Order Insights listing. Click on it.
   • If needed, set the Sort By to Most Recent or search for your account locator.
4. Click the blue Request Access button.
   • If prompted to verify your email address, click on the user profile link in the dialog to complete verification.
   • Verify the email sent to you. Then reload the browser tab.
5. In the Request access dialog, enter a business justification such as:
   • "We need this data for our current project xyz."
6. Submit the request.
7. After submitting, click the gray View request button to review your pending request. You can also withdraw and resubmit the request from here if needed.

1. Log in to your account as the provider user (switch browser tabs/windows or log in again).
2. Navigate to the Data Sharing -> Internal Sharing .
3. Open the Requests tab at the top of the Internal Sharing page.
4. You should see the access request from consumer. Click on it to review the details:
   • The requesting user and role
   • The business justification
   • The listing being requested
5. Click the green Grant button to approve the request.
6. Switch from Needs Review to Resolved to confirm the request is now listed as approved.

Switch back to the consumer user and navigate to the Order Insights listing in the Internal Marketplace. The blue Request Access button should now have changed to Query in Workspaces (reload the browser tab if needed). This confirms that access has been granted successfully.

Now that access has been granted, let's consume the Order Insights data product as the consumer user with the consumer_role.

1. Log in to your account as the consumer user (or switch browser windows).
2. Navigate to Catalog > Internal Marketplace.
3. Open the Order Insights listing. Reload the browser tab if you still see the "Request Access" button.
4. Click Query in Workspaces. Snowflake opens a new worksheet pre-populated with the sample queries from the listing.
5. Review and run the sample queries.
6. Keep the worksheet open for the consumer side of the next (optional) exercise, to use the same use database context.

A key benefit of organizational listings is that data is shared live. When the data owner updates the underlying data, consumers see the changes instantly. Let's demonstrate this.

1. Switch to the provider user and run the following in a worksheet:

   USE ROLE provider_role;
   USE WAREHOUSE compute_wh;
   USE SCHEMA tpch.sf1;
   
   -- Check current country for customer 60001
   SELECT customer_name, country, orderkey, orderdate, AMOUNT
   FROM TABLE(orders_per_customer(60001));
   

   Note the country for customer 60001 (Kenya).

2. Now update the customer's nation:

   -- Customer 60001 moves from Kenya to Mozambique
   UPDATE customer SET c_nationkey = 16 WHERE c_custkey = 60001;
   

3. Switch back to user consumer (e.g. switch browser tabs) and re-run the same query:

   USE ROLE consumer_role;
   
   SELECT customer_name, country, orderkey, orderdate, AMOUNT
   FROM TABLE(SF1.ORDERS_PER_CUSTOMER(60001));
   

   The updated country (Mozambique) is instantly visible to the consumer -- no refresh, no sync, no delay.

Now let's make your data product visible to the entire organization! In this lab, the organization consists of the Snowflake accounts of all the other lab participants around you.

You can do this manually by going back to the "Step 4 of 5: Configure Access Control and Approval" that you performed above. But, let's use Coco to do it for you.

1. Log in to your account as the provider user (switch browser tabs/windows or log in again).

2. Navigate to Data Sharing -> Internal Sharing .

3. Open the Listings tab at the top of the Internal Sharing page.

4. Select your live listing

5. Invoke Coco by clicking on the blue AI icon at the bottom right:

   

6. Prompt Coco to make the desired change for you:

   Please change the discoverability of this listing to the entire organization. 
   

   Then click Allow to confirm:

   

7. Hide Coco by clicking the ->| icon in the top right corner.

8. Reload the browser tab to validate the effect of Coco's ALTER statement in the UI:

   

Now let's look at the data products that others have published:

9. Switch to your consumer user.
10. Navigate to Catalog > Internal Marketplace. More and more data products published by other lab participants will appear in the internal marketplace! (Reload your browser tab if needed.)
11. Use the filters for Profile as well as for the custom attributes Confidentiality and Source Systems to display different subsets of data products.
12. Open at least 2 or 3 data products other than your own and Request Access as you did before for your own data product.
13. Navigate to the Data Sharing -> Internal Sharing -> Requests periodically to check if you have received any access requests from other lab participants. If so, approve them.

Let's make your data product AI-ready by generating and adding a semantic model (semantic view) and an AI agent.

1. Log in to your account as the provider user (or switch browser tabs/windows ).

2. Navigate to Data Sharing -> Internal Sharing .

3. Open the Listings tab at the top of the Internal Sharing page.

4. Select your listing.

5. Click on Secure share at the top:

   

6. You see the list of objects currently shared by your listing. Snowflake offers to generate and add a semantic view and an agent. Click Get started. Then click Create.

   

7. Click Try to test drive the semantic view with some natural language prompts. A new browser tab will open.

   • Sample prompt: "What questions can you help me answer?"
   

8. Return to the browser tab with your listing and click Add to secure share and Done.

9. Your data product is now marked as Cortex AI Ready.

1. Log in to your account as the consumer user (or switch browser tab).

2. Navigate to Catalog > Internal Marketplace.

3. Open your Order Insights listing.

   • Note that the blue "Open" button now provides multiple options
   • You can access the shared data directly or open the semantice view or the agent instead
     

4. From the "Open" button open the agent.

5. Click on "Preview in Snowflake Intelligence". A new tab opens and you may need to login in again as the data consumer.

6. Issue one or more prompts to talk to your data, such as:

   • What is the total number and value of sales per country?
   • Are there any outliers or anomalies in this data that I should be aware of?

7. Examine the results, charts, and analysis provided by Snowflake Intelligence.

   

Let's use Coco to create and manage listings through prompts.

Next:

Next, let's ask Coco to implement a new governance requirement.

Issue the following prompt:

Coco may ask you a series of questions to get this done.

Once Coco has created the policy, switch to the consumer user to see the effect of the masking, e.g. by selecting rows from the SUPPLIER table in a worksheet:

Congratulations, you completed this Snowflake Internal Marketplace hands-on lab! You have learned how data products are authored, published, requested, and consumed. You have made data products AI-ready by adding semantic views and AI agents, and you analyzed the shared data using natural language in Snowflake Intelligence. You also managed listings with Cortex Code.