A context engineer is building a multi-agent Databricks workflow where an orchestrating agent coordinates five sub-agents, each performing document analysis, entity extraction, and summarization tasks. The orchestrator receives the complete raw output from each sub-agent, including intermediate reasoning chains, extracted entity lists, and status messages, and appends all of it to its running context window before issuing follow-up instructions. After scaling to production document volumes, the orchestrator's context window consistently saturates before all sub-agent results are processed, causing truncation errors and a loss of task coherence.
Which two changes to sub-agent output design would reduce the orchestrator's context load while preserving the information it needs to continue coordinating tasks?
Choose two

• A. Merge all five sub-agents into a single agent so the orchestrator receives one combined output instead of five separate outputs per workflow run.
• B. Configure sub-agents to store full outputs in a Lakebase table and return only a record identifier to the orchestrator.
• C. Increase the orchestrator's model context window by switching to a larger foundation model so that all sub-agent outputs fit without truncation.
• D. Reduce the number of sub-agents so that fewer tokens are sent to the orchestrator before final synthesis.
• E. Restructure sub-agent outputs to return only a concise structured summary object containing task status, key extracted entities, and a confidence score, omitting intermediate reasoning chains.

A context engineer is creating a Databricks agent to triage incoming customer support tickets for a B2B software platform. Each ticket averages 200 tokens.
For each ticket, the agent must:
(1) assign a severity rating of P0/P1/P2/P3 based on whether the issue is fully blocking, degrades core functionality, affects a non-critical feature, or is informational-a judgment that requires reading the ticket body rather than matching keywords; (2) route the ticket to one of eight engineering teams based on the technical area implied by the ticket; (3) generate a brief, context-aware acknowledgment reply.
The pipeline must process up to 500 tickets per minute with an end-to-end latency budget of under 2 seconds per ticket. The token budget per invocation is 2,048 tokens, and the operations team has flagged that inference costs must remain minimal at this throughput.
Which reasoning mode should the context engineer configure for the agent?

• A. Reduced thinking, because the task requires modest structured reasoning across four severity tiers, a small thinking budget fits within the 2,048-token invocation, and latency and cost stay manageable at
  500 tickets per minute.
• B. Extended thinking with a default thinking budget, because severity classification across four tiers and routing across eight teams benefits from deep chain-of-thought reasoning to maximize accuracy.
• C. Extended thinking with max_tokens capped at 1,024, because limiting the response length reduces cost while preserving full reasoning depth.
• D. Standard mode with extended thinking disabled, because skipping all reasoning overhead minimizes both per-invocation latency and inference cost at 500 tickets per minute.

A context engineer runs a finance Q & A agent on Databricks. The agent invokes a Genie space backed by tables in the finance.silver schema, and Genie chooses which table to query for each user questions by reading Unity Catalog table and column metadata. Two tables exist in the schema: revenue and revenue_gross.
MLflow evaluation runs show the Correctness scorer flagging frequent failures in which Genie queries revenue_gross when the user asks about net revenue, and queries revenue when the user asks about gross figures. Both tables are in active production use and must remain available to the Genie space, and the same disambiguation should be usable by other Genie spaces and BI consumers building against the same schema.
Which intervention resolves this failure?

• A. Apply a Unity Catalog certified-status tag to revenue and a deprecated-status tag to revenue_gross so that downstream consumers treat revenue as the authoritative source.
• B. Apply a custom Unity Catalog tag categorizing one table as the primary revenue source and the other as a secondary view, so Genie's tool selection prefers the primary.
• C. Add a Genie space instruction at the space level stating which table to use for net-revenue questions and which to use for gross-revenue questions, so Genie applies the rule when routing each user question.
• D. Add COMMENT text on both tables and their columns in Unity Catalog that clearly distinguishes the two tables - what each represents, the units, and the basis, so Genie reads the disambiguating descriptions when routing each user question.

A context engineer is preparing to deploy a per-tenant Databricks agent for a SaaS retail analytics platform.
Each platform customer is provisioned with their own Unity Catalog catalog, retail_ & lt;customer & gt;, and within each tenant catalog, data is organized by business domain rather than by origin.
The sales schema contains the customer's POS transaction and returns tables, which the customer's source systems write to throughout the day, alongside the daily_sales_by_store, hourly_revenue_by_category, and top_skus_weekly summary tables that a nightly Lakeflow Jobs pipeline produces by reading those POS tables. The inventory schema contains stock-level and inbound-shipment tables written continuously by ingestion, alongside demand_forecast_7d and reorder_suggestions tables produced by the SaaS's ML pipelines. The customer schema contains customer master records and contact tables, all populated by ingestion.
The SaaS deploys one agent per customer through Model Serving, and each agent has its own service principal. Before any agent endpoint is registered, the governance requirements are that an agent for a given customer must be able to retrieve data only from that customer's catalog and never from another customer's catalog, and that within its own customer's catalog, the agent must be able to retrieve only from tables populated directly by the customer's ingestion pipelines, never from nightly summaries, forecasts, or other tables produced by SaaS pipelines.
Which governance approach should the context engineer implement for each tenant's agent before registration?

• A. For each tenant's agent service principal, grant USE CATALOG only on that tenant's retail_ < customer > catalog and, within that catalog, grant USE SCHEMA on each domain schema together with table-level SELECT only on the tables populated by the customer's ingestion pipelines. Leave the nightly summary, forecast, and other SaaS-produced tables in those schemas ungranted.
• B. For each tenant's agent service principal, grant USE CATALOG only on that tenant's retail_ < customer > catalog and USE SCHEMA together with SELECT on the sales, inventory, and customer schemas, relying on the per-tenant catalog boundary to scope retrieval to the customer's data.
• C. Provision a single agent service principal shared across all tenants because per-tenant service principals are operationally heavy at the SaaS's expected scale. Grant the shared service principal USE CATALOG on every retail_ < customer > catalog and table-level SELECT only on the ingestion-fed tables in each catalog. Configure the agent's retrieval tool to inject the requesting customer's catalog name into every query so the agent reads only the right tenant's tables at request time.
• D. For each tenant's agent service principal, grant USE CATALOG on that tenant's retail_ < customer > catalog and USE SCHEMA together with SELECT on the sales, inventory, and customer schemas.
  Apply a Unity Catalog tag source=ingest to every ingestion-fed table across the tenant catalogs and configure each agent's retrieval logic to issue queries only against tables carrying that tag.

A context engineer is debugging a Databricks agent that supports a multi-turn customer support application.
The agent keeps a running summary of all conversation history across every user session in a single vector index and retrieves it on every turn through semantic search. Users report that the agent loses track of facts stated earlier in the same session, while unrelated details from other users' past sessions appear unexpectedly.
Latency has also increased as conversation volume grows.
Which storage configuration correctly resolves the memory type mismatch that is causing this degraded behavior?

• A. Use a session-scoped short-term buffer to retain all within-session conversation turns, and reserve the vector store for durable cross-session user facts retrieved by semantic search.
• B. Partition the vector index by user ID and session ID so that semantic retrieval is scoped per session, preventing cross-session bleed while reserving long-term facts for all users.
• C. Use a session-scoped short-term buffer to retain all within-turn context, and write a full conversation summary to the vector index at the end of each turn so semantic retrieval stays current across all active sessions.
• D. Replace the vector index with a relational table partitioned by session ID, and retrieve all rows matching the current session on every turn to maintain full conversational context for both short-term and long-term memory needs.