Match Suggestions

Match suggestions use AI to automatically propose bindings between your variables and data points in a connector's data model. Instead of manually finding and mapping each variable, the platform analyzes names, data types, and context to suggest the best matches — then presents the results inline for you to review.

Prerequisites

Match suggestions require:

• A connector with at least one data model version (created by running a discovery crawl or uploading a data model manually)
• Unbound variables on assets attached to the connector

Running the matching

Open the connector detail page and go to the Bindings tab. Click the Run Matching button to start the AI matching process.

The matching job runs asynchronously in the background. While it is running, an inline progress banner appears at the top of the bindings tab showing:

• A progress bar with the current completion percentage
• The number of variables being analyzed

You can continue working in the platform while the matching runs. If you navigate away and return, the progress banner reappears automatically — matching state is not lost on page reload.

Reviewing suggestions

When the matching job completes, the suggestions appear inline next to each variable in the bindings tab. Each variable with a match suggestion shows a colored indicator based on the AI's confidence level.

Confidence levels

Suggestion details

Click on a suggestion indicator to expand the details. Each suggestion displays:

• Suggested address — the data point address from the connector's data model (for example, an OPC UA node ID or Modbus register)
• Confidence badge — color-coded confidence level
• Accept and Reject buttons — to act on the suggestion individually

Filtering by suggestions

Use the filter controls on the bindings tab to show only variables that have pending suggestions. This makes it easy to work through all suggestions without scrolling past already-bound variables.

Accepting and rejecting suggestions

You can act on suggestions individually or in bulk.

Individual actions

For each suggestion, click Accept to create a binding with the suggested address, or Reject to dismiss the suggestion. Accepted suggestions immediately create a binding for the variable.

Batch accept

Click Accept All High-Confidence to accept all suggestions with a high confidence level in one operation. The platform creates bindings for all accepted suggestions and displays a summary showing:

• Number of bindings created successfully
• Number of warnings (if any)
• Number of failures with error details

Suggestion lifecycle

Match suggestions are tied to the data model version that was active when the matching job ran. Suggestions follow these lifecycle rules:

• New data model version — when a new data model version is created (for example, by running a new discovery crawl), all existing suggestions from the previous version are automatically expired. This prevents stale suggestions from referencing outdated data points.
• Auto-rematch — if auto-rematch is enabled on the connector, a new matching job is automatically submitted when a new data model version is created. This keeps suggestions up to date without manual intervention.
• Daily cleanup — expired and rejected suggestions are cleaned up automatically by the platform.

When to use match suggestions

Match suggestions work best when:

• Variable names and data point names follow similar naming conventions (for example, active_power matches a data point named ActivePower)
• You have a large number of variables to bind (dozens or hundreds)
• The connector's data model has descriptive data point names

For small numbers of variables or when names do not correspond, manual binding may be more efficient.

Comparison with manual binding