August 2026 Summaries

Physical AI, which enables machines and industrial systems to sense, reason, and act in real-world conditions, requires both detailed historical data for training and real-time telemetry for inference, presenting challenges for traditional data historians that were not designed for the speed and data precision Physical AI demands. Traditional historians face four main gaps: limited real-time access, signal compression, IT/OT fragmentation, and localized site architectures. InfluxDB 3 addresses these issues by augmenting existing historian systems with a distributed layer that facilitates edge inference and cross-site training, allowing for more comprehensive data access and processing. This modern architecture supports real-time decision-making at the industrial edge while consolidating operational data across enterprise environments, thus bridging data silos without disrupting existing historian investments. By integrating telemetry with asset metadata and multimodal data types, InfluxDB 3 enhances the training of Physical AI models, enabling them to learn from context-rich datasets and respond effectively to dynamic operating conditions.