StackAdapt introduced Ivy Studio on July 28, 2026, a natural-language hub that folds planning, forecasting, analysis, optimization and execution into a single agent-driven surface inside its demand-side platform. The company disclosed that its own employees run more than 15,000 AI-assisted workflows each week through Ivy, the underlying engine, and that close to 70 percent of Ivy Studio activity clusters in just 20 tasks.

The announcement, issued from Toronto at 9:00 a.m. Eastern Daylight Time, marks the second act of a product line that began thirteen months earlier. StackAdapt launched Ivy, an in-platform assistant, on July 9, 2025, positioning it as a way for agencies and brands to speed up decisions during campaign management. At that point the tool handled questions: recommended CPMs for connected television buys, ad specifications for digital out-of-home and native formats. The company reported roughly 1,700 in-platform messages in the thirty days before that launch.

Ivy Studio raises the ambition considerably. Instead of answering questions inside an existing workflow, it proposes to become the workflow.

What the product does

Ivy Studio sits directly within the StackAdapt platform rather than alongside it as a separate console. According to StackAdapt, the hub covers planning, forecasting, analysis, optimization and execution, and marketers interact with it in natural language: discovering audiences, evaluating campaign performance, building forecasts, identifying opportunities and taking action from one interface. AI agents supply supporting context and recommendations as work proceeds.

The framing in the release is deliberate. Most advertising platforms, the company argues, are built around reactive workflows that require users to navigate dashboards, reports and interfaces before any meaningful action becomes possible. Ivy Studio inverts that sequence. Marketers state the outcome they want; agents grounded in what StackAdapt calls its advertising intelligence surface the insights and recommend the next steps.

StackAdapt also stated that it is using OpenAI's language models to develop bespoke capabilities purpose-built for advertising and marketing, extending Ivy Studio with experiences designed for advertiser workflows. The company did not specify which models, which components run on them, or how those calls interact with client campaign data.

That last point matters for buyers evaluating the product. The release describes transparency and control as retained throughout, but does not document what the control layer looks like in practice: whether agents require approval before executing changes, what audit trail exists, or where the boundary sits between recommendation and autonomous action.

The 15,000 figure, read carefully

Two numbers anchor the announcement, and they measure different things.

The first is that StackAdapt employees now use Ivy to power more than 15,000 AI-assisted workflows every week, spanning campaign management, reporting, troubleshooting and optimization. This is internal usage by the vendor's own staff, not customer adoption. It is a reasonable proxy for whether a tool survives contact with daily operational work, and it is a considerably harder number to manufacture than a customer count. It is not, however, evidence of external demand.

The second is that nearly 70 percent of Ivy Studio usage concentrates in the top 20 mission-critical workflows. StackAdapt presents this as reflecting strong adoption for the tasks that matter most. Read another way, the same figure describes a narrow surface: the bulk of activity runs through a small set of repeatable operations, which is what one would expect from a system whose early strength lies in routine reporting and diagnostics rather than open-ended strategic work.

Kelsey Carey, Ad Operation Supervisor, AI & Innovation at Colour Media, described the practical shape of that usage. "Ivy Studio streamlines campaign optimization by helping our team quickly identify and act on key performance drivers," Carey said, according to the announcement. "Its programmable skills allow us to pull routine metric reports across multiple campaigns into unified outputs, making it easier to analyze real-time insights and make data-driven decisions faster."

Routine metric reports across multiple campaigns. That is a concrete, verifiable use case, and it aligns with the concentration figure rather than contradicting it.

Vitaly Pecherskiy, Co-Founder and CEO of StackAdapt, set out the broader claim. "Ivy Studio changes the paradigm of how marketers use our platform. By moving core capabilities into an AI-first environment, we are enabling teams to accomplish in minutes what previously took hours or even the coordination of multiple teams. We are fundamentally shifting from operating software to driving business growth," Pecherskiy said.

A crowded eighteen months

Ivy Studio arrives at the end of a build-out cycle that has left almost no significant demand-side or supply-side platform without an agent story.

PubMatic launched AgenticOS with live campaigns running on January 5, 2026, with WPP Media, Butler/Till and MiQ among early participants. Yahoo DSP followed a day later, embedding agents that monitor, diagnose and execute corrective actions across planning, activation, optimization and measurement. The Trade Desk shipped Koa Agents with Stagwell as pilot partner on April 22, 2026, publishing the Open Agentic Kit as a proposed standard the same day. Skaiused its ShopAble conference on May 13, 2026, to claim the first agent-native marketing operating system. MiQ added a Planning Agent to Sigma in June, pairing strategy formation with the existing Trading Agent.

Two weeks before the Ivy Studio announcement, Adform opened 29 read-only skills against its FLOW DSP, a release explicitly scoped to querying rather than acting.

The architectural pattern across these launches is consistent: agents absorb monitoring-heavy and repetitive work while humans retain authority over consequential decisions. StackAdapt's language fits that template.

Where StackAdapt diverges

One difference is worth flagging, because it separates Ivy Studio from a large share of the surrounding activity.

Much of the agent infrastructure built during this period runs on the Model Context Protocol, the open standard Anthropic released and later donated to the Linux Foundation. Amazon Ads moved its MCP server into open beta at the IAB Annual Leadership Meeting on February 2, 2026, allowing external agents including Claude, ChatGPT and Gemini to reach its advertising APIs through a single integration. Meta opened write-access connectors in April 2026. Google extended background task handling for Gemini agents in July 2026. The IAB Tech Lab has been assembling interoperability standards since January 2026 precisely to prevent the fragmentation that competing proprietary protocols would produce, extending OpenRTB, AdCOM, OpenDirect, VAST and the Deal API. By March 2026 its Agent Registry held ten entries, all operating under MCP.

Ivy Studio, as described, is not that. It is built directly within the StackAdapt platform, and the announcement makes no reference to MCP, to the Ad Context Protocol, or to any mechanism by which an external agent might reach StackAdapt's capabilities. The value proposition rests on depth of domain grounding rather than breadth of connectivity: the company states that as AI agents become common, the advantage will come from how well those agents understand advertising, not simply from their existence.

That is a defensible position. It is also a closed one, and for agencies running buys across four or five platforms, the practical question is whether a well-grounded agent inside one system saves more time than a general-purpose agent that can reach all of them.

The adoption evidence beneath the launches

Survey data collected across 2026 suggests the gap between agent capability and agent trust remains wide.

Research from Taboola's advertiser platform, conducted in March 2026 across 200 marketing leaders in the United States and United Kingdom, found that 76 percent of senior performance marketers report meaningful improvements from agentic campaign tools while those benefits concentrate almost entirely inside search and social platforms. Adoption of AI-powered campaign solutions reached 98 percent for Performance Max and Advantage+ against 80 percent for the open web.

TripleLift's 2026 survey of 200 advertising professionals across six countries, published May 19, 2026, found 73 percent using AI for optimization, with creative remaining the least automated component of campaign execution.

Mediaocean's 2026 H2 Market Report, drawn from 312 marketing professionals surveyed in May 2026 and released June 16, identified AI media as the fastest-growing investment category at 60 percent, with AI adoption for data analysis at 50 percent and an orchestration gap holding at 86 percent importance. An Adobe study covered in July 2026 referenced January 2026 research finding marketers automate only 18 percent of daily tasks while estimating another 29 percent could be automated.

The IAB's 2026 Outlook Study, released January 28, 2026, forecast 9.5 percent year-over-year growth in US advertising spend, with two-thirds of surveyed advertisers focusing on agentic systems for autonomous campaign execution and cross-platform measurement rising to 72 percent priority status from 64 percent a year earlier.

Those figures describe demand. They do not describe delivered results, and the distinction has been the recurring weakness in agentic marketing across the sector.

StackAdapt's year

The Ivy Studio release continues a pattern of paired capability additions. StackAdapt unveiled a Snowflake Native Apppowered by Snowflake Cortex AI on July 24, 2025, becoming the first demand-side platform to embed Cortex AI within the Snowflake Native App Framework. It integrated iHeartMedia broadcast radio inventory on November 10, 2025, and enabled advance digital out-of-home booking through a Broadsign partnership the same month.

On January 6, 2026, the company brought self-serve National Provider Identifier targeting to pharmaceutical advertisers, with HIPAA-compliant impression-level reporting. In February it secured EU-U.S. Data Privacy Framework certification, a mechanism permitting lawful transfer of personal data from the European Union without Standard Contractual Clauses. That coverage put the platform's processing volume at 465 billion automated optimizations per second.

Then on May 5, 2026, StackAdapt joined OpenAI's ChatGPT advertising pilot as a technology partner, alongside Criteo, Kargo, Adobe and Pacvue. OpenAI opened its self-serve Ads Manager to all US businesses the same day, dropped the minimum spend threshold entirely and introduced cost-per-click bidding alongside its CPM model.

The relationship between those two OpenAI touchpoints is asymmetric and worth separating. In May, StackAdapt became a buyer of OpenAI inventory. In July, it disclosed that it is a consumer of OpenAI models. The first is a media supply arrangement. The second is an infrastructure dependency.

Why this matters

For media buyers, the operational claim is specific enough to test: work that consumed hours or required coordination across multiple teams compressing into minutes. Colour Media's stated use, consolidating routine metric reports across multiple campaigns into unified outputs, is the version of that claim most likely to hold, because reporting consolidation is a bounded problem with a verifiable output.

For agencies managing multi-platform buys, the interoperability question sits underneath everything. AI infrastructure dominated the industry agenda at Cannes Lions in June 2026, where model developers courted advertisers and agentic capability became the default vendor pitch. Every platform now offers a hub. Whether those hubs talk to one another determines how much of the promised time saving survives a real media plan.

For anyone evaluating vendor claims, the disclosure gaps in this announcement are the useful signal. StackAdapt did not publish pricing, general availability dates, geographic scope, the number of customers with access, whether Ivy Studio is generally available or in limited release, or how the 70 percent concentration figure was measured and over what period. The company named one customer. Those omissions are common across agentic product launches in 2026, and they are the reason the sector's adoption surveys keep finding enthusiasm running ahead of measured outcomes.

Timeline

Summary

Who: StackAdapt, the Toronto-based advertising and orchestration platform led by Co-Founder and CEO Vitaly Pecherskiy, introduced the product. Kelsey Carey, Ad Operation Supervisor, AI & Innovation at Colour Media, provided the sole named customer comment.

What: Ivy Studio, an AI-first hub built inside the StackAdapt platform covering planning, forecasting, analysis, optimization and execution through natural language. It runs on Ivy, the company's AI engine, and incorporates OpenAI language models for bespoke advertising capabilities. StackAdapt disclosed more than 15,000 internal AI-assisted workflows per week and nearly 70 percent of Ivy Studio usage concentrated in the top 20 workflows.

When: The announcement was issued on July 28, 2026, at 9:00 a.m. Eastern Daylight Time, thirteen months after the original Ivy assistant launched on July 9, 2025.

Where: Toronto, distributed through Business Wire. The platform operates globally across connected TV, digital out-of-home, display, native, audio and email inventory.

Why: The launch lands inside a sector-wide build-out of agentic infrastructure that has run since late 2025, with PubMatic, Yahoo DSP, The Trade Desk, Skai, MiQ and Adform all shipping agent products. StackAdapt's approach differs in being closed to external agent protocols, betting that domain grounding inside one platform beats cross-platform connectivity. Survey data through 2026 consistently shows advertiser enthusiasm for agentic systems running ahead of measured results, which makes the disclosure gaps in this announcement, covering pricing, availability, customer counts and measurement methodology, the practical limit on what buyers can currently verify.