mParticle Agent FAQ

General

Does the agent take action automatically?

No. The agent provides answers and prepares proposals for supported tasks. It doesn’t apply a proposed change until you explicitly confirm it.

What happens if the agent doesn’t understand my request?

It tells you what it could and couldn’t resolve, then asks a focused question or offers the closest useful next step. It won’t guess silently or invent event or attribute names.

What data can the mParticle Agent access, and how is it protected?

For detailed information about what data the mParticle Agent can access, how it handles and processes the data it can access, and the related privacy implications of the mParticle Agent, see Data and Privacy.

Audiences

What types of audiences can the agent create?

The mParticle Agent creates individual real-time audiences, including hybrid real-time audiences that use eligible existing Composable Audience membership. It cannot create Composable Audiences or multi-node audience groups.

Can the agent create or modify an audience group?

No. The agent can recommend an audience group when your business goal would be best achieved with a multi-node audience, but it can’t create or restructure audience groups. Use the Audience Builder and the Audience Group Editor when you need to create and manage audience groups.

Can the mParticle Agent create a hybrid audience?

The mParticle Agent cannot create a Composable Audience. It can use an eligible existing Composable Audience as membership in one individual real-time audience. When that membership is combined with live behavioral criteria, the result is a hybrid real-time audience.

Can the agent edit an existing audience?

No. The agent can review available information about a saved audience, but it can’t edit, replace, activate, deactivate, or delete an audience. Use the Audience Builder to make changes to saved audiences.

What if my request needs a calculated or predictive attribute?

The mParticle Agent can use an existing calculated or predictive attribute in a supported audience definition when the attribute is active and available to you. If the attribute is unavailable, the agent explains the dependency and directs you to the appropriate workflow. It cannot create calculated attributes or modify existing predictive attributes. For supported prediction-based audiences, see “Which predictions can the mParticle Agent use to create an audience?” below.

Does the agent support Match Boost, Household Reach, and Audience Expansion?

The mParticle Agent can’t apply these features while building an audience, but you can apply them through the standard Audiences workflow after saving.

Can the agent activate or connect an audience to an output?

No. After you save an audience, follow the standard workflow to activate it and connect it to an output. For instructions, see Connect an Audience.

Predictions

Which prediction types can the mParticle Agent create?

The mParticle Agent can create Future Behavior predictions after you review and confirm a proposal. It cannot create Similar Customers or Best Offering predictions; you must create those yourself in the Predictions workflow.

The agent can recommend any of the three prediction types for your business goal. It can also review saved configurations and interpret current usable results for all three.

What is Calculation Readiness?

Calculation Readiness is a pre-creation assessment of whether a completed Future Behavior definition appears likely to calculate. It can be Checking, Ready, At risk, Blocked, or Unverified. Calculation Readiness is separate from configuration completeness and pipeline capacity, and it does not guarantee a successful calculation or final Strength.

What happens while a Future Behavior prediction is calculating?

Calculation can take up to 24 hours. The prediction is not ready for targeting until it is Active and has current usable results. You can return to the prediction later to review its status and continue the workflow.

What happens if a prediction calculation fails?

A failed current run is not usable. Values from a previous successful calculation may remain on profiles, but they are stale and should not be treated as current results. The Agent can help you understand the current state and consider another supported approach, but it does not edit or retry a failed saved prediction.

Does the mParticle Agent choose a target range for me?

No. Review the prediction details and select the exact target range that fits your use case. The Agent can explain the current status, freshness, Strength, and other available quality information, but it does not recommend, choose, or change the target range.

Which predictions can the mParticle Agent use to create an audience?

The agent can create one new individual real-time audience from a Future Behavior or Similar Customers prediction that is Active and has current usable results. You must select the exact target range yourself before the agent prepares the audience, and confirm the proposal before it saves the audience.

The agent cannot create an individual audience from a Best Offering prediction through this workflow because Best Offering requires a multi-node audience. Connect and activate a saved audience through the standard Audiences workflow.

Next steps

To investigate the data available in your workspace, see Explore Your Data.

• Build Audiences: Learn how to prepare, review, and save an audience with the agent.
• Create Predictions: Learn how to create a Future Behavior prediction and interpret supported predictions.
• Data and Privacy: Understand the agent’s data access and privacy boundaries.