Data Analyst Associate · Domain 7: Developing, Sharing, and Maintaining AI/BI Genie Agents (Genie Spaces) · 12% of the exam
Task 7.2: Assign permissions via the UI and distribute Genie Agents (Genie spaces) using embedded links, external app integrations, and channel apps; Optimize AI/BI Genie Agents (Genie spaces) by tracking user questions, response accuracy, and feedback; updating instructions and trusted assets based on stakeholder input; validating accuracy with benchmarks; refreshing Unity Catalog metadata
Genie Agent permissions, embedding, apps and Slack, and improving an agent with monitoring, feedback, benchmarks, better metadata and refined instructions.
Study it
Sharing, embedding and tuning Genie Agents with benchmarks
Lesson coming
Sample questions
Try each one before opening the answer. Every option is explained, with the Databricks documentation page that proves it.
Question 1 · choose 1
The sales team wants @Genie mentions in its #sales Slack channel to be answered by the team's curated sales Genie Agent, not by general Genie One answers. A Slack workspace administrator has installed the Databricks Genie app for Slack. What should the team do next?
AHave each member pick the sales Genie Agent on the app's Home tab
BHave each member pick the sales Genie Agent as the scope in a personal channel configuration
CAsk a Databricks workspace admin to select the sales Genie Agent in the workspace's External access settings
DHave the channel owner pin the sales Genie Agent for the whole #sales channel
Show the answer and why
AHave each member pick the sales Genie Agent on the app's Home tab
Incorrect
The Genie scope chosen on the App Home applies to that user's 1:1 conversations with the app, not to @Genie mentions in a channel.
BHave each member pick the sales Genie Agent as the scope in a personal channel configuration
Incorrect
In a channel, only a channel-wide pin routes conversations to the agent; a personal configuration in a channel uses Genie One.
CAsk a Databricks workspace admin to select the sales Genie Agent in the workspace's External access settings
Incorrect
Those workspace settings decide whether the app may connect to the workspace; they do not choose the agent that answers in a channel.
DHave the channel owner pin the sales Genie Agent for the whole #sales channel
Correct
A channel owner can choose a specific agent as the Genie scope and save it as a channel-wide setting, which routes the channel's @Genie conversations to that agent.
The Databricks Genie app for Slack answers DMs and @Genie mentions with Genie One or with a Genie Agent pinned in a DM or channel. In a channel, the channel owner sets the Genie scope with /databricks-genie config and saves it as a channel-wide setting; a personal configuration in a channel still uses Genie One, and the scope picked on the App Home covers the user's 1:1 conversations.
Three Genie Agents use pg_crm.public.accounts, a table in a foreign catalog over a PostgreSQL source that supports comment syncing, with foreign comment fetching turned on. Last week the source team rewrote the column comments, but the agents still show the old descriptions. The owner wants all three agents to get the new comments. What should she do?
ANALYZE collects statistics for the optimizer; it does not refresh synced comments.
BRun DESCRIBE EXTENDED on the foreign table pg_crm.public.accounts
Correct
To refresh comments from a supported source without querying the table data, Databricks says to run DESCRIBE EXTENDED on the foreign table.
CEdit the column metadata in one agent's UI, which the other agents then reuse
Incorrect
Metadata edited in the Genie Agent UI applies only to the agent where it is written.
DRun REFRESH MATERIALIZED VIEW on pg_crm.public.accounts
Incorrect
The foreign table is not a materialized view, so there is nothing for a refresh to recompute.
Databricks can sync table and column comments from supported external data sources, and DESCRIBE EXTENDED on the foreign table refreshes them without querying the data. If comments are unavailable, workarounds are editing column metadata in a Genie Agent (it applies to that agent only) or a materialized view over the federated table that carries its own comments.
Finance analysts work in Microsoft Excel and want to ask natural-language questions about governed Databricks data and get the answers as rows and columns in their workbook. What do they use?
AGenie One from the Databricks Excel Add-in
BA notebook dashboard exported to Excel
CA Databricks SQL alert with an Excel attachment
DAn OpenSharing (Delta Sharing) share opened in Excel
Show the answer and why
AGenie One from the Databricks Excel Add-in
Correct
Genie One is available from the Databricks Excel Add-in and brings results into Excel as native rows and columns.
BA notebook dashboard exported to Excel
Incorrect
A notebook dashboard does not take natural-language questions in Excel.
CA Databricks SQL alert with an Excel attachment
Incorrect
Alerts notify on thresholds; they do not answer questions.
DAn OpenSharing (Delta Sharing) share opened in Excel
Incorrect
Sharing gives access to data, not a natural-language chat in Excel.
From the Databricks Excel Add-in, users open Genie One, ask questions about governed Databricks data in natural language, and bring tables of results into Excel as native rows and columns.
Account users open Genie One at the account level. What can they see there, and what do they need to view data from a specific workspace?
AEvery asset in every workspace; no entitlement is needed
BOnly workspace admins' assets; they need the Databricks SQL entitlement
COnly assets shared with them; Consumer access for that workspace
DOnly Genie Agents; workspace admin rights for the workspace
Show the answer and why
AEvery asset in every workspace; no entitlement is needed
Incorrect
Users can only access assets explicitly shared with them.
BOnly workspace admins' assets; they need the Databricks SQL entitlement
Incorrect
Users see assets shared with them, and the workspace requirement is the Consumer access entitlement.
COnly assets shared with them; Consumer access for that workspace
Correct
At the account level users access only assets explicitly shared with them, and viewing a workspace's data requires the Consumer access entitlement.
DOnly Genie Agents; workspace admin rights for the workspace
Incorrect
Genie One also lists dashboards, apps and other shared assets, and admin rights are not required.
Account-level Genie One gives account users a unified view across workspaces, but users can only access assets explicitly shared with them; to view data from a specific workspace, the Consumer access entitlement is required for that workspace.
A Genie Agent owner wants a quick summary of recent message volume, active users and thumbs up or down feedback. Where does she find it?
AIn the SQL warehouse monitoring page
BIn Catalog Explorer's table insights
CIn the agent's General instructions text block
DIn the Weekly digest of the Monitor tab
Show the answer and why
AIn the SQL warehouse monitoring page
Incorrect
Warehouse monitoring shows compute metrics, not agent feedback.
BIn Catalog Explorer's table insights
Incorrect
Table insights show table usage, not Genie Agent feedback.
CIn the agent's General instructions text block
Incorrect
Instructions guide answers; they do not report usage.
DIn the Weekly digest of the Monitor tab
Correct
The Weekly digest on the Monitor tab shows weekly message volume, active users and thumbs up/down feedback.
The Monitor tab of a Genie Agent shows questions and responses; its Weekly digest section summarizes weekly message volume, active users and thumbs up or down feedback.
A Genie Agent owner wants a report on common themes and issues in user messages and feedback from the last six months. What starts that analysis?
ARunning a benchmark
BAnalyze Agent Usage, which launches Genie Code
CDESCRIBE HISTORY on the agent's tables
DA Databricks SQL alert on the agent's message counts
Show the answer and why
ARunning a benchmark
Incorrect
Benchmarks test answers to set questions; they do not analyze six months of usage.
BAnalyze Agent Usage, which launches Genie Code
Correct
Analyze Agent Usage launches Genie Code to review the last six months of messages, feedback and issues and report common trends.
CDESCRIBE HISTORY on the agent's tables
Incorrect
Table history records writes to tables, not user messages.
DA Databricks SQL alert on the agent's message counts
Incorrect
Alerts check a query result against a threshold.
On the Monitor tab, Analyze Agent Usage launches Genie Code, which reviews the user messages, feedback and issues from the last six months and reports on common trends.