PR Newswire
GroundTruth has unveiled Dynamic Intent Prediction, an AI‑driven audience solution that leverages ZeroToOne.AI’s Large Behavioral Model to forecast who will act, what they will do, and when they will do it. The announcement marks the first large‑scale commercial deployment of the model and promises to reshape how brands allocate media budgets across programmatic, CTV, and retail media networks.
What’s New: Dynamic Intent Prediction
The New York‑based adtech firm introduced Dynamic Intent Prediction as a new class of audience segment that refreshes every 24 hours. Unlike traditional look‑alike lists that extrapolate from past conversions, the solution predicts in‑market intent across windows ranging from a single day to a month. According to the release, the highest‑confidence segments are eight to nine times more likely to convert than standard audiences, delivering a reach footprint roughly 50 × larger than conventional targeting.
How the Large Behavioral Model Works
ZeroToOne.AI’s Large Behavioral Model (LBM) is built on a transformer architecture similar to large language models, but its training objective is future human behavior rather than next‑word prediction. The model ingests over 15,000 behavioral signals from more than 7 million U.S. points of interest, retrains weekly, and scores billions of mobile advertising IDs (MAIDs) across 9,000+ categories and brands. By updating predictions at the ID level daily, the system can surface consumers who are “in‑market right now,” a capability the press release describes as moving from “guessing to predicting.”
Why It Matters for Advertisers
The shift from static, historically anchored segments to real‑time, behavior‑forward audiences has tangible financial implications. GroundTruth cites a global restaurant chain that identified $100 million in wasted media spend, saw a 27 % lift in store visits, and tripled app downloads after deploying the LBM. A leading automotive brand reported a 30 % increase in media efficiency, a 400 % surge in website traffic, and a financial‑services revenue rise. These case studies echo Gartner’s forecast that AI‑driven targeting will account for roughly 30 % of ad spend by 2027, underscoring the commercial relevance of predictive audiences.
Competitive Landscape
Dynamic Intent Prediction enters a crowded market where major platforms—Google, Amazon, and Meta—already offer AI‑enhanced audience solutions. However, most competitors rely on first‑party data or aggregated third‑party signals that refresh on a weekly or monthly cadence. GroundTruth’s daily refresh and the breadth of its 15,000+ signal set differentiate it from the more static offerings of traditional demand‑side platforms (DSPs). IDC projects the AI‑in‑adtech market to exceed $23 billion by 2026, suggesting ample room for niche innovators that can deliver higher conversion lift at lower waste.
Implications for Enterprise Marketing Teams
For marketers managing multi‑channel campaigns, the ability to activate audiences across CTV, OTT, mobile, digital out‑of‑home, and desktop from a single platform simplifies workflow and reduces reliance on disparate data providers. The release notes that the segments are available through GroundTruth’s self‑serve Ads Manager as well as managed‑service arrangements for agencies and large enterprises. This flexibility aligns with Adobe’s recent emphasis on unified customer profiles, enabling marketers to plug predictive intent directly into activation engines without extensive data engineering. Enterprise marketing teams benefit from a streamlined approach.
Industry Insight
Rosie O’Meara, GroundTruth’s senior vice president of product, emphasizes that “the industry has been describing every refinement of historical data as ‘predictive.’ It isn’t.” Her comment highlights a broader shift in adtech rhetoric: moving from descriptive analytics toward prescriptive, action‑oriented AI. Likewise, ZeroToOne.AI’s co‑founder Naseer Hashim frames the LBM as a platform‑level breakthrough, noting that “predicting human behavior is one of the hardest problems in AI, and the most consequential.” As more brands adopt real‑time intent signals, the competitive bar for audience accuracy will rise, pressuring legacy DSPs to accelerate their own model refresh cycles.
Market Landscape
The adtech sector is in the midst of an AI renaissance. Programmatic spend in the United States topped $120 billion in 2025, and Forrester predicts that 65 % of marketers will allocate a portion of their budget to AI‑driven audience solutions within the next two years. Privacy‑first regulations such as the CCPA and GDPR have pushed vendors toward first‑party and consent‑based data, making real‑time behavioral modeling a strategic necessity. GroundTruth’s Dynamic Intent Prediction, which relies on anonymized signal aggregation rather than personally identifiable information, positions it well amid tightening compliance standards.
Top Insights
- Real‑time intent outperforms look‑alikes: GroundTruth reports up to 9× higher conversion likelihood versus static audiences.
- Daily refresh cuts waste: A global restaurant chain reclaimed $100 M in wasted spend after adopting the Large Behavioral Model.
- Cross‑channel activation: Audiences can be deployed on CTV, OTT, mobile, DOOH, and desktop from a single platform, streamlining media buying.
- Competitive edge through breadth: Over 15,000 signals across 7 M locations give the model a richer context than most DSP‑native solutions.
- Regulatory resilience: The solution’s reliance on aggregated, consent‑driven data aligns with emerging privacy frameworks.
Get in touch with our Adtech experts
BusinessWire
