Last updated: September 16, 2026
Why this exists
Most of the industry still decides how to inspect and maintain equipment using methods that have not fundamentally changed in twenty years. Damage mechanism tables, generic failure frequencies, and fixed inspection intervals were codified when the data to do better did not exist. They have been carried forward largely unchanged since. They were never designed to use the volume of inspection, operating, and failure history that facilities now have.
Pinnacle took a different approach. We built statistical models on large datasets so that equipment life cycles could be forecast from evidence rather than from published rules, and so that inspection and mitigation strategies could be planned around what the data shows about each asset’s actual risk. That is how Newton produces different answers than the conventional methods, and why those answers can be checked against outcomes.
Those models have to be maintained. Degradation behavior varies across materials, services, and operating environments, and a model that was well calibrated for one population may not hold for another. The way to keep models sound, and to build new ones where the industry is still relying on rules of thumb, is to keep testing them against large, diverse datasets.
That is the reason Pinnacle creates de-identified and aggregated data from facilities around the world. Better models mean equipment failures are anticipated rather than discovered, inspection effort goes where the risk actually is, and the people who work around this equipment every day are safer for it. That is the commitment behind this work: to help the industry make better decisions about the reliability and safety of its assets, so that every customer benefits.
What we create
De-identified and aggregated data is derived from information processed in Newton after all identifying details have been removed. Pinnacle uses it to build benchmark datasets: large collections of inspection history, measured degradation, and equipment attributes drawn from many facilities and many customers. Pinnacle uses these datasets to:
- Test Newton’s statistical models against the conventional methods the industry uses today, by forecasting outcomes from historical data and checking the forecasts against what was actually measured
- Improve existing models and develop new ones where the results show room to do better
- Improve Newton’s features and default configurations
- Develop future products and services, improve industrial safety, and other valid business purposes.
Pinnacle retains the intellectual property rights in the de-identified benchmark datasets and aggregated data it creates from it.
How de-identification works
Before Pinnacle creates a benchmark dataset, it removes the information that could tie the source data back to a company or a site:
- Company names, site names, unit names, and any customer-assigned identifiers
- Equipment tags, measurement location identifiers, serial numbers, drawings, and free-text fields that could contain identifying details
What we will not do
- We will not disclose, publish, or share data that identifies a customer, a site, or an asset.
- We will not produce aggregated outputs from populations so small that a customer, site, or asset could be inferred from them.