Loading post…
DatabricksBradley Munday, Tyler HickeyExplainer
A multi agent approach audience intelligence
Summary
Advertisers often lose strategic intent when campaign briefs become data queries, while planners and data teams also miss insights hidden across large customer datasets. An AI-powered audience generation solution on the Databricks Data + AI Platform addresses this gap by combining Genie, custom tool-calling agents, Agent Bricks, Unity Catalog, and a Databricks App. Planners describe audiences in natural language; Genie produces governed SQL segments, an Affinity Agent calculates statistically validated lift against the full population, and a Supervisor Agent routes or chains these capabilities. The resulting workflow saves audiences in Unity Catalog for auditability and downstream activation, and is described as compressing planning and iteration cycles from days or weeks to minutes while enabling faster response to market shifts and pacing issues.
Context
The post addresses fragmentation between campaign strategy and data execution. Planners may rely on limited signals, while analysts lack full strategic context, causing strategy dilution, incomplete feedback, and data blind spots across datasets containing thousands of attributes and millions of customers.
Approach / What changed
The solution ingests first-party and partner data, uses Spark Declarative Pipelines to create a unified population attributes table in Unity Catalog, and places a curated Genie Space over that data. Genie converts natural-language audience requests into governed SQL, an Affinity Agent computes additional statistical patterns, and Agent Bricks routes or chains those agents through a Supervisor Agent. A Databricks App exposes the workflow, while saved Unity Catalog tables support auditability and activation through downstream channels.
Takeaways
Campaign briefs serve as the North Star for audience creation, data science modeling, and activation, but translating strategic intent into data terms can dilute important requirements.
The Affinity Agent analyzes a generated segment against the full population and reports statistically validated lift, such as cryptocurrency investment at 2.5x baseline or exceptional fitness commitment at 5.7x in the example.
Agent Bricks lets a Supervisor Agent route audience-building requests to Genie, affinity questions to the Affinity Agent, and complex requests through both agents in one conversation.
FAQ
How does the system keep natural-language audience creation transparent?
Genie translates a planner’s request into a precise query and provides the SQL logic behind the audience definition. Audiences are also saved as persisted tables in Unity Catalog for future viewing, management, and auditability.
What does the Affinity Agent add beyond a requested audience segment?
It analyzes additional behaviors and attributes to identify patterns that were not specified in the original brief. Its findings are presented as statistically validated lift calculations against the full population, rather than assumptions.
How does the Supervisor Agent handle multi-step requests?
It routes simple requests to the sub-agent best equipped to answer them. For a complex request, it can have Genie build the segment first and then send it to the Affinity Agent for analysis before returning a unified response.
Source and freshness
- Source
- databricks.com
- Published
- Added to Yomu
- Last updated
Summaries are generated from the original by a language model. Check the original before relying on it.
Keep reading
Related reading
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
The best engineering reads, once a week
One useful email. Unsubscribe anytime.