AirMettle AirStat

AirStat by AirMettle delivers the breakthrough solution for enabling interactive exploratory analytics on massive multi-dimensional data. Explore the true statistics of your entire dataset through an elegant, efficient way of capturing the forest from the trees.

Powered by our patent-pending AirTree technology (U.S. Patent Application 20250217930 A1), AirStat lets data engineers, scientists, and analysts create and explore rich, compact statistical fingerprints of enormous datasets in seconds.

Most teams already rely on probabilistic sketches for fast approximations on massive data. AirTree finally makes the same workflow exact and infinitely richer: it turns large 1–4D numerical arrays from files or distributed batches into compact, perfectly mergeable fingerprints that preserve every count with zero loss. Merge shards instantly with no approximation error. Query full-resolution CDFs, precise percentiles, actual top-k bins, or complex multi-dimensional grids directly on the fingerprint itself — all at trivial computational cost.

This transforms the exact questions analysts ask most into instant, interactive insights: “What is the true 95th percentile?”, “Show me the exact median.”, “What are the real top-k outliers?”, “What is the joint distribution inside this bounding box?”, or “How has the distribution evolved across shards or time windows?”

Key Benefits
• Capture the essential statistical character of massive multi-dimensional datasets at a tiny fraction of the original
scale — giving you an instantly queryable “forest-level” view while preserving perfect counts within every
discovered region.
•. Blazing speed with zero approximation trade-offs — unlike traditional probabilistic sketches, AirTree fingerprints
deliver exact answers to rich statistical queries at interactive speeds.
• Intelligent discovery → precise follow-up — rapidly identify regions of key interest, then issue simple, highly selective
queries against the original data to retrieve only the exact tiny fraction you need.
• Perfect merging and incremental updates — combine AirTrees from any number of shards or time windows with
zero loss.
• Dramatically lower cost and complexity — achieve interactive exploratory analytics on petabyte-scale data
with minimal data movement on Azure.

Key Capabilities
• Rich support for 1D through 4D AirTrees — compact, perfectly accurate statistical fingerprints purpose-built
for multi-dimensional numerical data
• Blistering interactive queries directly on AirTrees: precise percentiles, full-resolution CDFs, Top-K bins, min/max,
bounding-box filters, and multi-dimensional grid analysis
• Exact, lossless merging of any number of AirTrees from shards, batches, or time periods
• Export to Apache Arrow, Parquet, and CSV for seamless integration with your existing tools and pipelines
• Production-ready access via powerful CLI, high-performance C++ library with Python bindings coming soon

Ideal For
• Observability, telemetry, and monitoring platforms — high-cardinality metrics and logs generate enormous volumes
in object storage; AirTrees enable teams to instantly understand distributions, identify anomalous regions, and target
precise follow-up queries without scanning raw petabyte-scale data.
• Scientific computing and HPC simulation workloads — researchers routinely generate massive multi-dimensional
floating-point arrays from parallel jobs; AirTrees provide rapid exploration, exact merging across shards, and
compact storage while preserving full accuracy for validation and analysis.
• IoT, sensor, and industrial data pipelines — multi-dimensional streams (time × location × multiple sensors) demand
fast statistical and spatial insights; AirTrees deliver the lightweight fingerprints needed for real-time monitoring and
efficient drill-down into raw records.
• Financial services and risk analytics teams — precise, auditable distributional analysis of prices, exposures, returns,
and portfolio metrics at scale is critical; AirTrees give exact top-k, percentile, and joint-distribution views with
minimal compute, enabling faster risk decisions and compliance workflows.

AirStat turns your existing Azure object storage into a high-performance statistical intelligence layer with minimal friction. Fully compatible with Microsoft Fabric and OneLake workflows — AirStat processes data directly from Azure Blob Storage and ADLS Gen2 (including OneLake locations via shortcuts) and exports enriched results in standard open formats that are easily consumed in Lakehouses, Notebooks, and other Fabric experiences.

Start accelerating your data exploration and reducing storage and compute costs today.