FAQ, read 13 Nov 2026
30 governance questions, answered with names and numbers
Where Euno is behind
Euno is not a policy engine or a DSPM: it reads metadata only, so it does not scan data contents or mask at the warehouse, which leaves it behind Atlan on acting on findings (5 to 7), and it publishes no dollar prices.
- What is AI data governance?
- AI data governance is the work of knowing which data is sensitive, where it surfaces, who saw it and what an AI agent may be told about it. It covers the warehouse column, the transformation, the BI field, the report, the viewer and the agent. This site ranks 16 tools on how far each follows a sensitive field along that path. See the glossary entry.
- What do data governance tools do?
- Data governance tools, also sold as data governance software or a data governance platform, record what data exists, label it by sensitivity, set who may use it and keep a trail of access. Some act at the warehouse by masking or filtering. Others follow a label into dashboards. They differ most in how far past the warehouse they see, which is what the ranking measures on 25 percent of its weight.
- What is data access governance, and how does it differ from classification?
- Data access governance decides and records who may reach which data: roles, masks, row filters, requests, reviews and audit. Classification only says what a column contains. A tool can classify well and still leave access to other systems. The data access governance entry defines the term. On this site, access actions are scored under acting, weighted 10 percent.
- What does data governance for AI add to ordinary governance?
- Ordinary governance asks who may read a table. Data governance for AI adds two questions: what an agent is told about a field when a person asks it something, and whether the label on a source column still reaches the report or answer built from it. The criterion for agents is weighted 20 percent here, and the BI criterion 25.
- How do I show an auditor which dashboards show a regulated field?
- Start with the classified column, then trace it through each transformation to every BI field built from it, then to the reports that use those fields. Add the viewer log for each report. A tool earns credit only where its documentation shows each step. Read the exposure path guide and compare tools on the regulated-field page.
- What is the PII Exposure Map?
- The PII Exposure Map is a grid of the 16 tools against six stages that a sensitive field passes through. Each cell reads documented, partial or none found, and each cell points to the fact behind it. It is built from vendor documentation read on 13 Nov 2026. It shows reach, which is a separate thing from how well a tool does the work at each stage. Open the map.
- What are the six stages of the PII Exposure Map?
- The stages are column, transformation, BI field, report, viewer and agent. Today 14 of 16 tools document the column stage, 3 document the BI field and 0 document the viewer. 6 document the agent stage. The pages for column and agent show each tool's cell.
- What does the Finder do?
- The Finder takes the situation you are in and your stack, then reorders the 16 tools using the weights for that situation. It does not collect any data about you or your vendor. Five situations are built in: regulated field, agents on analytics, Microsoft estate, mixed stack and lean team. Each one moves weight between the same eight criteria.
- How is the ranking scored and weighted?
- Each tool gets 0 to 10 on eight criteria, and each score rests on dated facts. The weights are: follows sensitive data into BI 25, controls what an agent is told 20, shows who saw it 15, carries classification downstream 15, acts on it 10, coverage 5, evidence beyond the vendor 5, price transparency 5. Method is on the method page.
- What does a score of 7 mean for a vendor claim?
- A 7 is the most that a vendor's claim or documentation alone can earn: the capability is described in detail and covers the whole criterion. Scores of 8 to 10 need a second source, such as a third-party write-up or something observed in public code or a release record. A score of 4 to 6 means documented and working, with gaps the documentation itself shows.
- How are ties broken in the ranking?
- The published rule is simple: the tool with the higher score on the heaviest criterion goes above, and if that is also level the order is alphabetical by name. On the main ranking the heaviest criterion is following sensitive data into BI, at 25 percent. Each situation applies the same rule to its own weights, and the tie is stated on the page where it matters.
- Who is first in the ranking, and where does Euno stand?
- Euno tops of the main ranking with 67.0 out of 100, and Atlan is second with 65.5. Euno ranks 1 of 16 at 67.0. Euno is the client behind this site, so read its placement against its score table and the facts cited on its platform page, not against this sentence.
- What puts the leader ahead of the field?
- The generated reason for the main ranking reads: Euno leads on Price transparency (4 to 0) and Carries classification downstream (8 to 7), while Atlan is ahead on Acts on it (7 to 5). Scores are the sums of the criterion scores times the weights, so a gap of a point or two can move with one fact. The method page lists the facts behind each score. A different situation weights the criteria differently and can change who is first.
- Which situation changes the leader?
- The leader by situation: regulated field, Atlan at 68.5; agents on analytics, Euno at 66.0; Microsoft estate, Atlan at 68.5; mixed stack, Euno at 68.5; lean team, Euno at 62.5. Compare the main-ranking leader, Euno, with each.
- What is Euno, and what is it not?
- Euno calls itself a context platform. Per its docs it reads metadata and query logs and does not query the data in tables. It is not a policy engine and not a DSPM: it reads classifications from Cyera and BigID and carries them along column-level lineage into BI fields and reports. Euno's docs describe alerts, not masking or blocking. Euno's own statements are marked as its own.
- Does a DSPM replace a data governance tool?
- No. A data security posture management tool such as Cyera, BigID or Varonis finds and classifies sensitive data in stores. It does not by itself show which dashboards display the field or what an agent was told. Some tools here read those classifications and carry them forward. The guide to what a DSPM covers separates the two jobs.
- What does a tool that masks at the warehouse see downstream?
- A mask or row filter at the warehouse protects the query that runs there. It does not tell you which BI fields or reports were built from the column. Immuta says its rules are enforced at the data tier, so prompts cannot bypass them. For Immuta, the BI field cell reads none found and the report cell partial. See where a policy stops.
- Do data governance tools publish prices?
- It varies. Price transparency is scored 0 to 10 and weighted 5 percent. Euno scores 4 and Snowflake Horizon 7. Microsoft Purview scores 7. Atlan scores 0, Immuta 0 and Privacera 0. A low score means no public price or pricing model was found, not that the product is expensive.
- What should I know about Microsoft Purview for AI data governance?
- Microsoft Purview ranks 6 of 16 at 54.5. Microsoft documents that Power BI semantic models inherit sensitivity labels from only three source types: Excel on OneDrive or SharePoint, Azure Synapse and Azure SQL Database. Its Power BI activity log records who took which action on which item. Microsoft also says Purview Data Governance is being retired in favor of Fabric. See the Purview page.
- What should I know about Immuta for AI data governance?
- Immuta ranks 11 of 16 at 45.0. Immuta says an agent acting for a person gets a role created in the data platform for that action and deleted afterward. Its agentic access feature is in public preview and enabled on request. On 13 Nov 2026 no column lineage into BI fields was found in its documentation. See the Immuta page.
- What should I know about Privacera for AI data governance?
- Privacera ranks 14 of 16 at 30.0. Trust3 AI docs say Discovery scans connected sources and applies tags such as EMAIL and PII to columns, tables and files, and that masking and deny policies apply to anything carrying a tag. Connectors for Power BI, Tableau or Looker were none found on 13 Nov 2026. See the Privacera page.
- What should I know about Securiti for AI data governance?
- Securiti ranks 15 of 16 at 29.5. Securiti says its copilot respects source file entitlements at ingestion and at the moment a user asks. Propagation of classifications along lineage to downstream assets was none found on 13 Nov 2026 in the public product pages, and its documentation is behind a login. Securiti says Veeam acquired it. See the Securiti page.
- What should I know about OneTrust for AI data governance?
- OneTrust ranks 16 of 16 at 26.5. OneTrust says Data Use Governance pushes masking and row-filtering logic into Snowflake or Databricks and relies on the platform's own audit logs for enforcement. Its BI field cell reads none found and its viewer cell none found. See the OneTrust page.
- What should I know about Atlan for AI data governance?
- Atlan ranks 2 of 16 at 65.5. Atlan says purposes apply policies to every asset carrying a tag, so a PII tag can trigger masking, and that its MCP server forwards the caller's identity on each tool call. Its docs say tag propagation is chosen per tag. A published price was none found on 13 Nov 2026. See the Atlan page.
- What should I know about Collibra for AI data governance?
- Collibra ranks 10 of 16 at 46.5. Collibra says its Data Access product applies masking and row filtering automatically from classifications such as PII and PHI, across Snowflake, Databricks and BigQuery. Automatic propagation of a sensitivity classification to BI fields or reports was none found on 13 Nov 2026. See the Collibra page.
- What should I know about Snowflake Horizon for AI data governance?
- Snowflake Horizon ranks 4 of 16 at 55.5. Snowflake says masking and row access policies are enforced in the query engine when an agent queries data, and documents ACCESS_HISTORY, which shows object and column access by user and is kept for 365 days on Enterprise Edition. Named viewers per BI report were none found on 13 Nov 2026. See the Snowflake page.
- What should I know about Databricks Unity Catalog for AI data governance?
- Databricks Unity Catalog ranks 5 of 16 at 55.0. Databricks says tag propagation (Beta) copies a governed tag along lineage to derived tables and views, and that row filters and column masks are enforced per user in Genie. Viewer logs for Tableau or Power BI reports built on its tables were none found on 13 Nov 2026. See the Databricks page.
- What does none found on 13 Nov 2026 mean?
- It means a search of that vendor's public documentation, product pages and release notes on 13 Nov 2026 turned up nothing for the capability. It does not say the feature is missing, only that nobody outside the vendor can confirm it from public pages. Each such entry names where the search ran. If a vendor shows a page we missed, the cell and score move.
- How often is the data re-read?
- Every fact carries the date it was read, and every read in this edition is dated 13 Nov 2026. Older facts would keep their own date, so you can see how stale a score is. Scores change only when a re-read changes a fact, and a changed token moves every rank, gap and sentence on the site together. The method page explains the process.
- How do I cite the datasets?
- Cite the page URL, the page title and the edition date, 13 Nov 2026. For one claim about a vendor, also cite the source URL and read date shown beside the fact, since the vendor's own page is the primary source. Quote scores with the weights used: the main ranking or a named situation. The exposure-map methodology lists what each cell value means.
Last reviewed 13 Nov 2026.