Hi, Thank you for joining our webinar, Virtual Product Premiere - Spring ‘26 release. We hope you found the product announcements insightful. Questions / Answers Q1: For unstructured data cataloguing, many clients have file volumes far above the asset allowance, for example millions of files versus an allowance of around 1.5 million assets. How does Collibra recommend handling this at scale? A1: Collibra Platform can scale well beyond 1.5M assets and so larger asset volumes are supported. It is true that Deasy enables governance of a much broader array of unstructured data assets to be governed and Collibra Platform supports that as we would any other asset type. We're also looking to expand the integration of Deasy and the main Collibra platform to further simplify that integration. Additional asset bundles can be purchased as an add on through the normal sales process. Q2: A key part of context for structured data is a logical data model that shows how tables and entities relate to each other, for example customers and orders. How do you see this being addressed? How do you see Collibra supporting integration with AI agents in practice? We currently use APIs for this, but is Collibra planning a more direct connector or stronger integration pattern in the future? A2: Regarding logical / semantic context for unstructured data - Collibra allows for building/documenting logical models in the platform. By governing unstructured data in Collibra, we can map across physical and logical layers. Regarding giving agents access to Collibra context and metadata - beyond APIs, we are enhancing our MCP Server offering. Further integration options are under review. Q3: When wil Deasy Labs support OneDrive scanning? A3: Hi, Deasy supports cloud-based data across systems including OneDrive, SharePoint, and S3 to name a few. Q4: Does this work with Snowflake Cortex Ai, etc? A4: Hi, I'm seeing your questions a few minute late so don't recall the specific slide/subject. We don't have a native Cortex integration currently, but it is on our roadmap for 2026. Depending on the use case we can talk about if/how we would approach working with this. Q5: Co Pilot is GA now? A5: Collibra Copilot is currently in Public Preview, with GA timeline to be confirmed. Q6: Is Data Access an add on component or part of the standard release? A6: Hi, Data Access will be an add-on to the core platform. Q7: Can Data Access read in the groups managed by my Identity Access Management tool? I'm particularly interested in Active Directory, but there are others. A7: Thank you for your question. We're initially supporting Entra ID and Okta. In our July release you can read groups from your Identity Provider. In the August release you can actively manage groups in the Identity Provider with Data Access. Q8: Which of these features we are seeing today are widely avaialble versus private preview? A8: Great question. For the current capabilities shown, Semantic Mapping and Model Generation Agents are Generally Available. Control Tower is in Private Preview and we encourage you to join! In addition, AI-powered writing assistant is currently in Private Preview. Our Innovation Page provides more info: https://www.collibra.com/products/innovations Q9: Can Collibra Data Access integrate with other 3rd party tools/legacy? A9: Thank you for your question. We're initially supporting Databricks, Snowflake and BiqQuery, with others to come. We're also planning an SDK that lets you build your own connectors. Q10: How many CU consumption does this take as your adding the CU consuption now? A10: Hi - Deasy Labs does not use CUs for consumption. Q11: How many CU consumptions does this take? A11: Hi. For questions around Collibra Units, please reach out to your Account Representative. Q12: I missed the start of the session; will there be a recording shared? A12: Yes. We'll share in a follow up email. Q13: What's the difference between a data owner and producer. Who's accountable in this contract model approach? A13: Data Product Owner is the business person accountable for the data — they define what the data should represent, its quality standards, and its business value. Data Producer is typically the data engineer responsible for ensuring the data is actually built and delivered to meet the agreed specifications. We view the data contract as the formal agreement between these two parties. In short: the owner is accountable, the producer is responsible, and the contract is what ties them together. Q14: I think you are missing the value part to data products. Show real $ value of the data to the company/users. Show how access/products equates to value. A14: Great point, and we completely agree. This is one of our primary use cases ("Delivering ROI with data products"). We are releasing a private preview of our data product application, which will include a dedicated dashboard to track data product adoption and associated usage requests. This is a first step toward making the connection between data access and business value much more tangible. Q15: Need to be able to see the model of the trust score. A15: Not shown in the demo, but it is available in the settings for the Trust Score and you can define the weight of each metric. Also, there is a different model for business assets (use case) and technical assets (models, agents, etc). Q16: How will Data Access work with Snowflake? We currently provision groups with access and add people to those groups fro snowflake access. Authentication is done using Entra. A16: Collibra Data Access will manage access in Snowflake using Roles. As of July, we give Identities access by giving them access to Roles in Snowflake. As of August, we can also give Identities access by adding Identities to the Entra ID or Okta Groups that correspond with the Role (through SCIM integration). Q17: Are data contracts similar to policy, where we create a policy attach to Data asset. A17: Data contracts and policies are related but serve different purposes. A policy is a broad rule or standard. A data contract is much more specific — it is a formal agreement tied to a particular data product that defines exactly what is being delivered, by whom, and to what standard (schema, quality, SLAs). Q18: The AI code capture how does that work do I need to use API or ??? A18: This works through the Collibra CLI. Q19: Does it allow high volume users like etl users to be excluded from the masking SF native wont do this alone? A19: Yes. With Collibra Data Access you can manage global and scoped exceptions on data masks. This allows you to exclude the service accounts used in ETL from the data mask. Q20: Is data masking implemented via say the DBs native capabilities like snowflake masking databricks etc? A20: Thank you for your question. Yes, it is. Q21: For the semantic model does it use the BG for the labels or if not how can it conform ompeting tech lables in multiple tables? A21: Yes. When you run the Semantic Model Generation Agent, it cross-references your raw column names, existing column descriptions, and data classifications against the terminology defined in your Business Glossary (BG). Q22: Can we register the MCP servers also? A22: We do not (yet) have MCP as an asset type, it is possible to make a custom asset type for this, but we are exploring the option to include this. Q23: I assume with access it is working with Azure AD and can be done at the user/group levels? A23: Yes. There will be several ways of managing access. As of July, we give Identities access directly in the Data Warehouse. As of August, we can also give Identities access by adding Identities to the Entra ID or Okta Groups.