← All patterns Martech Patterns · November 18, 2026

Marketers Get a Better View Before Approving AI Changes

Analysis by the CDP Institute

Agents on marketing platforms now build audiences, adjust bids and set up activations, with a marketer approving each change or a set scope of work. That approval is built into the workflow: the agent shows why it proposed each change and logs every step it takes, so a marketer can check the reasoning before signing off and trace what ran afterwards.

  1. Dinmo launches an agentic CDP that uses warehouse data to prepare audiences and proposed actions

    What happened: DinMo launched an agentic customer data platform that runs on a company's own data warehouse. Its agent, Dan, investigates changes, builds predictions and prepares audiences, journeys and activations. Each change is approved individually or within a scope the team authorizes in advance.

    Why it's important: Teams can hand an agent the analysis and setup without letting it act unchecked. Every step it takes, including what it read and why, is recorded in an audit log, so a reviewer can trace how a campaign was built. Sending still happens in the team's existing tools.

    Source: dinmo.com

  2. AppsFlyer adds four MCP capabilities for marketing data and campaign management

    What happened: AppsFlyer expanded Agent Hub with customised workflows and an Apple Ads agent that applies approved bid changes. Four new MCP capabilities will cover fraud data, creative performance, campaign link creation and partner integration settings through existing AI tools.

    Why it's important: Marketers will soon be able to delegate analysis and campaign adjustments without giving up approval authority. Visibility into recommendation reasoning provides an additional basis for reviewing automated decisions before execution.

    Source: Communicate Online

  3. Basis broadens its advertising AI capabilities to include intelligence, automation and governance

    What happened: Basis expanded Compass with historical campaign intelligence, Morning Consult research, DSP activation and diagnostics, and audience creation through Dstillery’s DS-1. An MCP server also gives customers read/write access to Basis features through custom agents in Claude, ChatGPT or other tools.

    Why it's important: Media teams can connect planning decisions to campaign execution without moving among separate research and buying systems. Governed workflows and permissions provide control over AI access and actions.

    Source: basis.com

From the CDP Institute: What is a CDP? · Vendor Directory · News archive