AppsFlyer has unveiled a new AI-powered orchestration protocol that redefines how marketing teams access, analyse and act on campaign data. The new tool, known as Model Context Protocol (MCP), connects AppsFlyer’s attribution and analytics infrastructure directly to large language models (LLMs), enabling marketers to generate real-time insights and automate decisions through natural language queries or autonomous agents.
The goal is not to replace marketers but to remove the friction between performance data and execution. As campaign complexity grows, spanning platforms, audiences, and regulatory constraints, traditional workflows have struggled to keep pace. Most require specialist skills, rigid dashboards, or technical support teams to unlock critical insights.
Barak Witkowski, Chief Product Officer at AppsFlyer, says the system addresses this directly. “By connecting our attribution and analytics infrastructure to conversational AI, we’re reducing the friction between insight and action. This aligns with wider trends towards more intuitive, AI-enabled tools that empower marketers to move faster and smarter.”
AI agents transform campaign operations
At its core, the MCP product acts as an orchestration layer between AppsFlyer’s data APIs and LLMs such as Claude, enabling marketers to directly query metrics like lifetime value, retention, or audience overlap without navigating traditional interfaces. Teams can now generate targeted insights on demand, whether through human interaction or agentic AI tools acting in the background.
This dual-channel architecture allows for tailored outputs based on user roles. A chief marketing officer might ask for regional ROI breakdowns, while an analyst could request influence attribution trends for specific channels. MCP ensures each user receives a context-specific response without the need for shared training, eliminating long wait times and reducing dependency on specialist teams.
The protocol also addresses structural inefficiencies across the marketing tech stack. By surfacing real-time campaign performance, managing audience segmentation, and enforcing link governance through a single conversational interface, MCP offers a path to integrating AI deeper into the marketing decision-making layer. This shift toward embedded, autonomous intelligence reflects broader movements across enterprise functions, where AI is no longer a bolt-on enhancement but a fundamental interface layer.
Bringing inference to the front line
The significance of the announcement lies not only in AppsFlyer’s technology but in what it signals for the wider AI infrastructure conversation. While much of the industry remains focused on model training, AppsFlyer’s MCP is part of a wave of tools optimised for inference, the stage where AI delivers value through real-time responses and context-sensitive execution.
By anchoring these capabilities in high-quality, privacy-safe datasets, AppsFlyer is also addressing a growing enterprise concern: trust in AI-generated decisions. The integration ensures that each insight is grounded in verifiable data, reducing the risk of hallucinations and compliance breaches at a time when marketers are under mounting pressure to reconcile personalisation with privacy.
Early users, such as the mobile gaming rewards platform BUFF, report significant reductions in time-to-insight and a more distributed approach to data access. Where once entire teams waited on a handful of analysts, decision-makers can now access tailored information in minutes through simple queries, allowing the organisation to move at the speed of the market.
AppsFlyer MCP is now available through Claude, with plans to expand to additional LLMs and deeper platform functionalities. The company’s long-term roadmap suggests a vision of marketing not as a set of manual tasks or batch campaigns, but as an environment where AI agents operate alongside humans in real time, optimising, segmenting, and acting without manual triggers.
In a sector defined by split-second decisions and customer-level granularity, the shift to AI-native orchestration may prove more than a productivity boost. It could redefine the interface between insight and execution—turning every campaign into a living, adaptive system.
