GrowthLoop Inc. has launched a composable artificial intelligence analytics platform designed to understand the causal drivers of customer behavior.
The Composable AI Decisioning platform operates directly on enterprise data clouds, allowing marketers to analyze and act on customer data in near real time without moving it to separate systems. This enables businesses to direct messages to customers as they browse a website or engage on the phone.
According to former CEO Chris O’Neill, the platform integrates with major data clouds like Google BigQuery and Databricks, operating on top of existing data infrastructure. This contrasts with traditional methods that require data to be moved and hydrated into separate tools.
The launch signals a shift in marketing technology from correlation-based analytics to systems that establish causality, a capability enhanced by artificial intelligence. GrowthLoop's platform utilizes AI techniques, including reinforcement learning and multi-armed bandits, to expose consumers to different stimuli and test outcomes in a controlled manner.
Key features include near real-time operation without a separate customer data cache, achieved through technologies like Kafka queues. This enables “same-session personalization,” allowing marketers to adjust offers and messaging while a customer is actively engaged.
The system also offers continuous measurement capabilities, moving beyond static control groups to constantly snapshot interactions and integrate them back into the database for ongoing strategy refinement.
An “agentic context graph” component integrates multiple data sources to inform decision-making. While operating directly on cloud data platforms can introduce latency, GrowthLoop claims its technology mitigates these issues, eliminating round-tripping latency problems.
Audit controls are in place to address concerns about AI model behavior in automated decisioning, with every interaction snapshotted back into the data cloud.
O’Neill positioned the platform as part of a broader shift from linear campaign models to iterative learning systems, emphasizing “loops, not funnels” for continuous improvement through a partnership between AI agents and humans.