Predactiv is expanding its advertising data platform with the Predactiv People Model, an AI foundation model designed to create, analyze and activate audiences using behavioral signals. The model combines Predactiv’s proprietary open-web behavioral data with additional datasets to generate audience models, predictive insights and activation-ready segments.
The company says the model processes more than 65 billion unique behaviors each month, creating a behavioral foundation for custom audience creation and predictive modeling.
Market Landscape
Audience targeting is moving toward increasingly dynamic models as advertisers contend with fragmented data environments, privacy restrictions and the declining usefulness of static audience segments. The IAB’s 2026 State of Data report identifies data fragmentation, signal loss and the need for stronger data infrastructure as continuing challenges for marketers and media buyers.
Predactiv’s approach shifts audience creation from selecting predefined segments toward describing a desired audience and allowing an AI system to construct a segment from observed behavioral patterns.
The distinction is relevant to programmatic advertising because audience definition increasingly influences activation across DSPs, publishers and data platforms. A model capable of generating audiences dynamically could potentially reduce dependence on fixed third-party audience packages, although effectiveness ultimately depends on data quality, permissions and the downstream activation environment.
How the Predactiv People Model Works
The model generates an embedding for every user, according to Predactiv. These representations are intended to remove some of the feature-engineering work traditionally required for predictive audience modeling.
Advertisers can describe an audience using natural language, such as consumers likely to purchase premium running shoes within a specified period. The system then builds a custom audience based on behavioral signals.
Predactiv says the model can also support lookalike modeling, next-purchase prediction and custom modeling. The company claims these models can be created and deployed in minutes rather than weeks, though those performance claims have not been independently validated.
The platform also supports first-party data onboarding. Data owners can compare their customer populations with broader behavioral patterns and expand seed audiences into modeled audiences for activation.
Another component is natural-language analysis. Teams can ask questions about audience behavior and receive text-based answers derived from the model’s underlying behavioral data.
Programmatic and Enterprise Impact
For advertisers and agencies, the technology could provide an alternative to purchasing standardized audience segments. Instead, media teams can specify an audience objective or provide a seed audience and generate a segment tailored to a particular campaign.
Publishers and data owners could use the same infrastructure to enrich first-party datasets and develop additional audience products. For DSPs and other buying environments, the critical question will be how these modeled audiences translate into scalable activation while maintaining data permissions and interoperability.
Predactiv has made the People Model available through its Data Platform, Platform API and MCP Server. The MCP implementation allows teams to build, size and activate audiences from AI tools including ChatGPT and Claude.
Strategic Outlook
The combination of foundation models and audience intelligence points toward a more conversational model of advertising data. Instead of navigating predefined taxonomies, marketers could increasingly express business objectives in natural language and have AI systems translate those requirements into usable audience definitions.
That could make audience creation more accessible to marketers while also changing how data providers differentiate their products. The competitive value will increasingly depend not only on the size of a dataset but on the uniqueness of the signals, model quality, governance and ability to activate audiences across media environments.
Predactiv’s model therefore represents a shift from audience marketplace selection toward AI-assisted audience generation, with first-party data, behavioral modeling and activation increasingly connected within one workflow.
Top Insights
- AI-generated audiences: Marketers can describe desired audiences in natural language.
- Behavioral foundation: Predactiv says the model draws on more than 65 billion unique behaviors monthly.
- Predictive modeling: Embeddings support lookalike, next-purchase and custom audience models.
- First-party data expansion: Data owners can transform seed audiences into modeled, activation-ready segments.
- AI-native activation: The MCP Server connects audience workflows with AI tools such as ChatGPT and Claude.
Get in touch with our Adtech experts
