Datadog Launches Open-Source AI Tools for Observability Forecasting

Datadog has introduced two major artificial intelligence research projects aimed at reshaping how observability forecasting is approached. Unveiled on May 21, 2025, the company's new initiatives—Toto and BOOM—mark a significant step in open-source AI development for real-time telemetry data analysis.
Toto, a foundation model purpose-built for time series forecasting, sets itself apart as the first open-source tool of its kind focused specifically on observability data. Drawing from Datadog's extensive internal telemetry systems, the model provides zero-shot forecasting, enabling it to detect anomalies and predict infrastructure capacity without prior retraining or fine-tuning. Designed for flexibility and rapid deployment, Toto offers practical value in use cases where speed and adaptability are crucial.
Also read: Catena Labs Raises $18M to Build First AI-Native Financial Institution
Introducing the Largest Observability Benchmark
Complementing the model is BOOM, a public benchmark dataset for evaluating performance on observability tasks. With over 350 million observations across 2,807 real-world multivariate time series, BOOM offers unprecedented scale. The dataset encapsulates the nuances of production telemetry, including sparse data points, traffic spikes, and cold-start scenarios—conditions often overlooked in traditional benchmarks.
Together, Toto and BOOM offer a comprehensive framework for both developers and researchers aiming to improve observability tools. By open-sourcing both the model and dataset under a permissive license, Datadog is inviting broader community collaboration to refine forecasting models in enterprise settings.
Toward Practical AI in Infrastructure Monitoring
These launches represent a new phase in Datadog’s AI strategy. Rather than focusing solely on internal innovation, the company is now building bridges to the wider AI and research ecosystem. By making advanced tools accessible, Datadog hopes to accelerate the translation of academic breakthroughs into real-world operational improvements for infrastructure teams.
Both Toto and BOOM are immediately available for public download. Their release signals Datadog's growing investment in open research and its intention to make AI a foundational layer in the next generation of observability and monitoring solutions.