18 Levels of system integration
| Level | Access & Action | Description |
|---|---|---|
| 1. Historical Analysis (Passive Insight) | Data only – no machine or data stream access | Work from static CSVs or logs, providing insights, trends, or recommendations based solely on historical data. |
| 2. Monitoring & Warning (Observational) | Real-time data stream, no control | Ingest live sensor or data feeds; support threshold alerts and anomaly detection but cannot affect machine behavior. |
| 3. Feedback Loop (Latent Influence) | Network access; non-real-time control | Read from and write to systems remotely; can propose settings or adjustments (e.g., batch recalibration) but not apply instantly. |
| 4. Real-Time Autonomous Control (Fully Integrated) | Full local access and control | Compute and apply decisions in real time. Includes online feature processing, low-latency inference, even continual learning. |
18.1 Historical Analysis
The most basic level of system integration involves working with historical data that has been collected and stored over time. It may provide insights, trends, or recommendations based solely on this data, but it lacks the ability to interact with live systems or influence real-time operations.
E.g. analyzing historical production data to identify trends in equipment failures and recommending maintenance schedules.
18.2 Monitoring & Warning
The second level of system integration focuses on monitoring live data streams. Data science products at this level can analyze real-time data to identify potential issues or deviations from expected behavior, supporting operators in maintaining system performance and reliability.
E.g. detecting anomalies in the machine’s behaviour as it runs and alerting operators to investigate potential system failures.
18.3 Feedback Loop
The third level of system integration introduces a feedback loop, allowing for more dynamic interactions with the system. It is not possible to achieve real-time control or immediate adjustments, but it is possible to adjust machine settings or parameters based on insights gained from the data.
E.g. adjust machine parameters such as pressure, temperature, flow-rate for the next production batch based on an analysis of failures in post-inspection of previous batches.
18.4 Real-Time Autonomous Control
The fourth level of system integration enables real-time autonomous control of machines and processes. At this level, data science products can make decisions and take actions in real time, allowing for immediate adjustments and optimizations.
E.g. automatically adjusting machine settings in real time based on live data inputs to optimize production efficiency.