Trade Desk, Dstillery Use AI Agents to Reshape Audiences

The Trade Desk and audience intelligence firm Dstillery have announced an integration that connects their respective AI systems to reshape audience segments while a campaign is live, rather than at the planning stage. The move is a notable step in the broader industry push to make programmatic buying more autonomous, shifting audience construction from a pre-flight setup task to a continuous, in-flight optimization loop.

For a Tier 1 demand-side platform like The Trade Desk, this is more than a feature announcement. It signals how large buy-side platforms intend to deploy agentic AI inside the live auction environment — a direction that carries real implications for how publishers, SSPs, and video inventory are valued and matched to demand.

What the integration actually does

Traditionally, audience targeting in programmatic follows a linear path: a media buyer defines segments, maps them to data providers, loads them into the DSP, and launches. Mid-campaign adjustments are possible but manual — a trader reviews performance, identifies underperforming segments, and swaps in new ones. That cadence is measured in days, not milliseconds.

The Trade Desk–Dstillery arrangement aims to compress that loop. Dstillery's custom audience modeling — which builds predictive segments from behavioral and contextual signals — is connected to AI agents that can evaluate campaign performance and trigger the construction of new or refined audiences without a human manually rebuilding them. In practice, the system can detect which signals are driving conversions and expand, contract, or reshape the addressable pool accordingly, feeding those updated segments back into the DSP's bidding logic.

The key change is agency: software is making the targeting decision and acting on it, with the human moving into an oversight role rather than an execution one.

Why this matters for the programmatic stack

Audience definitions ultimately express themselves as bidding behavior. When a segment changes, the set of impressions a DSP is willing to bid on — and how aggressively — changes with it. If audiences are being reshaped continuously by AI agents, bid patterns on the supply side become more fluid and harder to predict from a publisher's or SSP's vantage point.

This has several downstream effects worth watching:

• Bid density and win rates may fluctuate more within a single flight as the targeting pool is reshaped, altering the competitive dynamics on any given impression.
• Yield predictability for publishers could become noisier, since demand from a given buyer is no longer anchored to a fixed, pre-defined segment.
• Frequency and reach planning shifts from a static model to a moving target, which has implications for pacing and budget delivery across display and video.

For CTV and video specifically, where impression supply is comparatively scarce and expensive, more precise and adaptive audience targeting could improve the efficiency of spend on premium inventory — but it also raises the bar for measurement. Attribution and viewability signals need to feed the optimization loop fast enough to be useful, which puts pressure on real-time data pipelines and identity resolution in a post-third-party-cookie environment.

The agentic AI trend line

This announcement fits a clear pattern across the buy side. The Trade Desk has been positioning its platform — including its Kokai interface — around AI-driven decisioning, and connecting external data partners' models to that system extends the reach of autonomous optimization. Dstillery brings specialized audience-building capability; The Trade Desk brings scale and auction access. The result is a division of labor between AI systems rather than between humans.

The open question for the ecosystem is transparency. Autonomous audience reshaping is powerful, but it can also make campaign behavior harder to audit. Ad ops teams and publishers will want clarity on which signals drive segment changes, how brand-safety and suitability constraints are enforced inside the loop, and whether identity inputs comply with privacy frameworks as targeting becomes more dynamic.

What to watch next

Expect other DSPs and audience vendors to pursue similar agent-to-agent integrations, and expect SSPs to respond with their own supply-side optimization logic to keep pace with increasingly adaptive demand. The practical test will be whether mid-campaign reshaping demonstrably improves outcomes — measured in conversion lift, effective CPM efficiency, and reduced waste — without degrading transparency or inflating costs.

For ad ops practitioners, the near-term takeaway is to understand where human control points remain: budget caps, suitability rules, and measurement guardrails become the primary levers when audience construction itself is automated. As agentic systems move deeper into the live auction, governance — not segment selection — becomes the core operational discipline.

Stay on top of video ad serving and programmatic. Follow Adelerate.