• Written by Azhar Aliyar, Director

Some people object to starting with integration for AsscherAi by saying it is too complex and time-consuming, but this is the wrong objection because a first thirty days on integration can produce something checkable if done correctly. The real challenge is in choosing the right approach and understanding the mechanisms involved. A shift supervisor checking the maintenance log, for instance, may find that the data from the maintenance system does not match the data from the purchasing system, which can lead to incorrect conclusions about the state of the equipment.

Choosing a first case

Choosing a first case that already has a known answer to compare against is crucial for getting started with AI systems integration. This allows for verification of the integration process and identification of any potential issues. For example, a category manager comparing two suppliers can use the purchase order table to verify the data from the supplier management system. Most people get wrong the idea that any data will do, but they are wrong because the quality of the data and the complexity of the case can greatly affect the outcome.

A good first case should have a clear and simple answer, such as the total cost of a specific project or the number of employees in a particular department. This makes it easier to verify the results and build confidence in the integration process. A simple case also allows for a more focused approach, reducing the risk of getting bogged down in complex data issues.

Read paths and write paths

Read paths and write paths are two critical components of integration, but they require different levels of care. Read paths are relatively straightforward, as they only involve retrieving data from a system. Write paths, on the other hand, involve updating or inserting data into a system, which requires more care to avoid data corruption or inconsistencies. A failure to update the data correctly, for instance, can cause the purchase order table to become outdated, leading to incorrect decisions.

The failure to update the data correctly is often caused by a mismatch between the data formats of the two systems. For example, the purchasing system may use a different date format than the supplier management system, which can cause errors when trying to update the data. To avoid this, it is essential to carefully plan and test the write paths to ensure that the data is updated correctly and consistently.

Reconciling definitions

Reconciling definitions that differ between systems is a common challenge in integration. Different systems may use different terminology or data formats to describe the same thing, which can lead to inconsistencies and errors. For instance, the maintenance system may use the term "equipment" to refer to a specific type of machinery, while the purchasing system uses the term "asset".

To reconcile these differences, it is essential to carefully map the data fields and definitions between the two systems. This involves identifying the equivalent fields and definitions in each system and creating a mapping that allows the data to be translated correctly. This process can be time-consuming and requires careful attention to detail, but it is essential for ensuring that the data is accurate and consistent.

What to stop doing if the comparison fails

If the comparison between the integrated data and the known answer fails, it is essential to stop and re-evaluate the integration process. This may involve re-checking the data mappings, verifying the data formats, and testing the write paths. It is also important to check for any data corruption or inconsistencies that may have occurred during the integration process.

A common mistake is to try to force the data to match by making assumptions or adjustments to the data. However, this can lead to further errors and inconsistencies, and can undermine the integrity of the data. Instead, it is better to take a step back and re-evaluate the integration process to identify the root cause of the issue. For more information on troubleshooting integration issues, visit our support page.

When integration is not the answer

Integration is not always the answer, and there are cases where it may not be the best approach. For instance, if the data is highly sensitive or subject to strict regulatory requirements, it may be better to use a different approach, such as data encryption or access controls. Additionally, if the systems are highly customized or have complex data structures, integration may not be feasible or may require significant resources and expertise.

In such cases, it may be better to consider alternative approaches, such as data migration or system replacement. It is also important to consider the costs and benefits of integration and to weigh these against the potential risks and challenges. For more information on whether integration is right for your organization, contact us at contact-us.