We are excited to announce the first phase of NLP-Based Segmentation, a core component of our new Agentic AI suite. This feature bridges the gap between natural-language ideas and technical segment configurations, significantly reducing the time required to create audiences.

Overview

The Segment Agent allows users to define complex audience rules using simple English prompts. Instead of manually navigating dropdowns and logic gates, you can now describe who you want to target, and the AI will build the rule-set for you.

• Natural Language Input: Define segments like: “Users who have opened any email in the last 30 days” or “Customers who clicked a push notification at least twice in the past week.”

• Direct Integration: Once the AI generates the segment logic, it can be directly appended to your existing Segment Builder for further refinement or immediate use in a campaign.

Workflow

• Prompt: Enter your segment criteria into the input field.

• Generate: The Segment Agent interprets the intent and maps it to your specific event data (e.g., Email Opens, Click events).

Review: Preview the generated logic and click to add it directly to your Segment Builder.

Support & Fallback Protocol

As this is a beta release, we are continuously training the model on complex query logic.

• In case of failure: If the agent is unable to accurately translate a prompt, users will be prompted to use the Manual Segment Builder to ensure no interruption to campaign schedules.

• Feedback Loop: Users are encouraged to “Raise a Query” via support for any failed prompts. Our engineering team logs these instances to refine the NLP accuracy for the upcoming release.

Key Use Cases for Clients

• Rapid Retargeting: Quickly identifying users who interacted with a specific channel in a recent timeframe.

• Behavioural Bucketing: Segmenting users based on frequency of actions (e.g., “did X more than 3 times”).

• Efficiency for New Users: Allowing team members who are less familiar with the technical schema to build accurate audiences without extensive training.

Refer to our documentation for more information.