Join our session “What AI Agents Need Before They Touch a Media Decision” at Advertising Week New York on Wednesday, October 7, to see the AI Gateway for Advertising on the Tech Stage, sponsored by Snowflake.
Across marketing, AI is moving away from being just a tool that assists and is instead becoming an agent that acts, helping teams diagnose performance, decide what to do next and run while campaigns are live.
Advertising is one of the sharpest proving grounds for that shift. Yet campaign teams still move between ad platforms, pacing sheets, Slack threads and raw data to understand performance and reconcile conflicting metrics. The problem isn’t just fragmented workflows; it’s fragmented context. An agent may see a platform’s metrics without the business definitions or channel signals it needs to know what matters, leading to a recommendation that misses a core KPI or doesn’t fit the channel.
Snowflake is introducing the AI Gateway for Advertising, currently in preview, with governed Model Context Protocol (MCP) integrations for Meta and TikTok, with PMG as an early customer. Connected to Cortex Agents, the AI Gateway brings together enterprise context, including business definitions and historical performance, with live platform signals such as delivery, catalog health and diagnostics. Agents can use the AI Gateway to reason across the full picture, recommend the next best optimization and implement under policy, with campaign managers approving and observing each step.
The AI Gateway for Advertising, built on Snowflake’s May 2026 acquisition of Natoma’s enterprise MCP platform, enables MCP integrations with Meta and TikTok. These connections, with additional platforms coming soon, supercharge Cortex Agents to run the full campaign loop: The agent reasons over cross-channel performance, diagnoses live status and settings, decides the best action, explains the rationale to a campaign manager in the loop and acts with permission.
An agent can combine live platform signals, such as delivery, catalog health and conversion diagnostics, with data that lives in the advertiser’s Snowflake environment, such as transaction history, inventory, and the business’s definitions of return on ad spend (ROAS) and conversion. The ad platform shows what is happening in the channel, and connecting it back to data and insights in Snowflake can help articulate whether it matters to the business. Together, they give the agent the evidence and context to explain a performance change and recommend the next action, all on a shared set of definitions and KPIs.
Let’s say a retail marketer sees ROAS drop on a live seasonal campaign. Spend is pacing correctly, and click-through rate (CTR) looks healthy. With no obvious flags, finding the cause for the drop becomes a multiday investigation that involves the ad platform, data team, catalog warnings and business data in Snowflake.
With the AI Gateway for Advertising, the marketer simply asks why. The agent combines campaign performance from the ad platform MCP with transactions, inventory and conversion-signal health in Snowflake to identify several contributing factors:
Purchase event quality declined after a checkout payload change
Several promoted SKUs have warnings
A high-spend product group has limited inventory
It recommends a bounded action: Reduce the affected prospecting ad set’s budget by 20%, but don’t pause it, as the converting cohort has strong lifetime value (LTV).
The recommendation, supporting evidence and approval are captured in one shared record, so the next person reviewing the campaign can pick up where the investigation left off.
In this example, a media marketer sees a strong response to an email promoting a new series, with one audience segment driving unusually high viewing and subscription activity. Turning that owned-media win into a paid-media action requires connecting email engagement, viewing behavior, subscription data, audience overlap and live campaign performance.
The marketer asks: Where should we increase paid-media exposure to build on this momentum? The AI Gateway for Advertising connects owned-media signals in Snowflake with live campaign data through the ad platform MCP and identifies that:
Viewers who clicked the email and watched the first episode are converting to subscriptions at a higher rate
This segment is underrepresented in the current paid audience
Its level is below the one associated with the strongest subscription lift
It would increase reach but dilute efficiency
It recommends a bounded action: Increase ad frequency for the email-engaged, Episode 1 viewer segment by 50% for the next seven days. Keep broader prospecting unchanged while the segment’s subscription and retention performance continues to validate the strategy.
The audience signal, rationale and approval are captured in one shared record, creating a repeatable handoff from owned-media insight to paid-media action.
Join our session “What AI Agents Need Before They Touch a Media Decision” at Advertising Week New York on Wednesday, […]
Credit: Robert Triggs / Android Authority TL;DR Samsung has refreshed its SmartThings app on mobile and smart home appliances. The […]
Observability doesn’t end when the incident ends Consider a hypothetical large financial services company serving millions of customers. Like many […]
Our services help both the big and small organisations in the identification of security breaches and provide outclass solutions to prevent any potential harm.