> ## Documentation Index
> Fetch the complete documentation index at: https://agenticbanking.backbase.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Map endpoints and fields
> Use AI-assisted endpoint mapping and schema mapping in Connector Studio.
This page covers the mapping work inside a connector project: endpoint mapping, schema mapping, and the Mapping Workspace.
## The two levels of mapping
Connector Studio maps at two levels, and you complete them in order. The following table describes each level:
| Level | What you match | Where you do it | How you start it |
| :--------------- | :--------------------------------------------------------------------------------------- | :------------------------------------------------------ | :------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Endpoint mapping | A source operation to a target operation, such as `GET /accounts` to `GET /arrangements` | The **Mappings** tab, in the **AI Suggestions** section | Usually starts automatically when you open **Mapping Workspace**. Select **Run AI Suggestions** if mapping didn't start, or if you want to run it again. |
| Schema mapping | A source field to a target field, inside one operation | The **Mapping Workspace** | Usually starts automatically after endpoint mapping completes. Select **Generate mappings** if mapping didn't start, or if you want to run it again. |
The two levels form a hierarchy:
* An API holds operations, and each operation is one endpoint. You map these first.
* An operation holds properties, which are its fields, enumerations, and transformations. You map these second.
Approve your endpoints first, then approve your properties. Schema mapping can only run for an operation after you approve or receive a suggestion for its endpoint pair.
The interface labels field-level work **Property Mappings**, while the backend and the AI messages call the same work **schema mapping**. Both names describe the same step.
## Actions at the top of a project
The following table describes the actions in the project header:
| Action | What it does |
| :----------------- | :------------------------------------------------------------------- |
| Mapping Workspace | Opens the detailed view of endpoint and property mappings |
| Route Designer | Opens the Camel route designer, if your role allows it |
| Download code | Downloads the generated connector artifacts |
| Run AI Suggestions | Asks AI to propose endpoint mappings. Set the target protocol first. |
| Analyse | Analyzes the codebase of a project you imported from GitHub |
## Project metrics
The metrics band on the project overview gives you a quick health check:
| Metric | What it reports |
| :------------- | :----------------------------------------------------- |
| Coverage | The percentage of source operations that have mappings |
| Approved | The mappings you accepted |
| Draft | The AI suggestions you haven't reviewed yet |
| Rejected | The mappings you declined |
| Avg Confidence | The average confidence score of the AI suggestions |
| Gaps | The source operations that have no mapping |
## Project tabs
The following table describes each tab on the project overview:
| Tab | What you use it for |
| :------------- | :---------------------------------------------------------------------------------- |
| Details | Edit the name, the organizations, the runtime configuration, and the Java constants |
| Mappings | Review and approve the AI endpoint mappings |
| Routes | See the generated Camel routes |
| Code Lists | Maintain reusable enumeration tables |
| Generated Code | Browse and download the output |
| Error Mappings | Map error codes, when your project enables them |
## Map endpoints
Start here after you create a project. Endpoint mapping usually starts automatically when you open **Mapping Workspace**. If it doesn't, or if you want to run it again, select **Run AI Suggestions**.
Open the project and select **Mapping Workspace**. A progress banner reports that **Endpoint Matching** or **Endpoint Mapping** is running when the job starts.
If endpoint mapping did not start, or if you want to run it again, select **Run AI Suggestions**.
On the **Mappings** tab, compare each suggested pair against the vendor documentation, then approve or reject it.
### What the Mappings tab shows
The tab groups endpoint pairs into two sections:
* **Approved Mappings** holds the pairs you accepted. Each row reports how many attributes, or properties, the operation contains.
* **AI Suggestions** holds the draft matches that AI isn't confident about. To hide low-confidence rows, use the **Confidence threshold** slider. To narrow the list further, filter by **Role**: Primary, Lookup, Enrichment, or Conditional.
### Actions on an endpoint row
The following table describes when to use each row action:
| Action | When to use it |
| :----------------- | :------------------------------------------------------- |
| Approve | The source and target operations are the correct pair |
| Reject | The match is wrong |
| Revert | You want to move an approved row back to the suggestions |
| Edit | You want to correct the paths or the metadata by hand |
| Run schema mapping | You want to map the fields for that endpoint only |
Do not approve rows or open the Mapping Workspace while endpoint mapping runs. If you see *Endpoint mapping is in progress* or *Schema mapping is in progress*, wait for the job to finish. The two workflows are not safe to run at the same time.
## Map schemas
Schema mapping answers one question: for this source JSON or XML structure, which target field corresponds to each source field?
### How schema mapping starts
Schema mapping starts in three ways:
| Trigger | What happens |
| :----------------------- | :----------------------------------------------------------------------------------------------------------------------------- |
| Automatically | After endpoint mapping finishes, endpoints with a confidence of about 60% or higher can start schema mapping in the background |
| For one endpoint | Select **Run schema mapping** on a row in the **Mappings** tab |
| In the Mapping Workspace | Select **Generate mappings** for the operation you're viewing |
While schema mapping runs, you see a **Property mapping in progress** banner, or a **Mapping schemas…** spinner on the row.
Do not approve property rows until the generation finishes. Rows can still change while the job runs.
## Use the Mapping Workspace
To open the workspace, select **Mapping Workspace** in the project header, or **Open Workspace** on the **Mappings** tab.
### Workspace metrics
The following table describes the metrics in the workspace header:
| Metric | What it reports |
| :---------- | :--------------------------------------------------------- |
| X% complete | The share of property work you approved, against the total |
| N approved | The property mappings you accepted |
| N suggested | The AI draft property mappings |
| N gaps | The properties that have no mapping yet |
### Select an operation
To move between endpoint pairs, use the operation switcher, which shows each pair as a source method and path followed by a target method and path. The switcher groups the pairs into **Approved**, **Partial review**, **AI Suggestions**, and **API Gaps**.
Before you map properties, select either **Request** or **Response**. Map each side separately.
### Map a field
The workspace has three columns:
| Column | Label | What it holds |
| :----- | :---------------- | :----------------------------------------------- |
| Left | SOURCE | The tree of fields from the source specification |
| Center | Property Mappings | The approved and suggested field pairs |
| Right | TARGET | The tree of fields from the target specification |
To create a mapping by hand:
In the **SOURCE** column, select the field you want to map.
In the **TARGET** column, select the field it maps to.
A new row appears in the center column, usually as an AI suggestion. Approve it to confirm the mapping.
If the vendor documentation is incomplete but you know the path from your own integration knowledge, select **+ Add manual property** under **SOURCE** or **TARGET**.
### Review property mappings
**Approved Mappings** holds the field pairs you trust. To change the strategy, the null handling, or the enumeration mappings, expand a row.
**AI Suggestions** holds draft field pairs, each with a confidence score. Approve a pair, reject it, or expand it to make a correction before you approve it.
The workspace header also has bulk actions:
| Action | What it does |
| :---------------------- | :---------------------------------------------- |
| Generate mappings | Runs AI schema mapping again for this operation |
| Approve All Suggestions | Accepts every draft property mapping at once |
| Unapprove All | Moves every approved property back to draft |
| Export | Exports the mapping data |
**Approve All Suggestions** accepts low-confidence pairs along with high-confidence ones. Raise the confidence threshold first, or review the pairs one by one, so that a wrong pair doesn't reach your generated transformation.
### Mapping strategies
Each property mapping uses a strategy. The following table describes when to use each one:
| Strategy | Use it when you want to |
| :------------------ | :------------------------------------------------------ |
| `direct` | Copy the value without changing it |
| `transform` | Apply a string, date, or number operation |
| `lookup` | Convert the value through a **Code List** |
| `constant` | Write the same fixed value every time |
| `conditional` | Produce a different result based on a rule |
| `split` or `concat` | Split one string into several, or join several into one |
| `complex` | Apply multi-step logic |
For the `lookup` strategy, create the table on the project's **Code Lists** tab first, then link to it from the property editor.
## Manual mapping
You can map endpoints and fields without waiting for AI suggestions.
### Manual endpoint mapping
On the **Mappings** tab, select **Edit** on an endpoint row to correct paths or metadata by hand. You can also create a mapping pair when AI didn't suggest one.
### Manual schema mapping
When a field is missing from the uploaded specification but you know the path from your integration knowledge, select **+ Add manual property** under **SOURCE** or **TARGET** in the **Mapping Workspace**.
## Gaps and confidence
A gap is a source property that has no target mapping. To close a gap, select **Generate mappings**, accept an AI suggestion, or map the field by hand.
Low confidence doesn't mean the suggestion is wrong. It means you need to verify it.
If Connector Studio indexed a specification only in part, you see **Partial schema index**. Upload the specification again, or wait for indexing to finish on the **API Specs** page, before you rely on automatic mapping.
## Troubleshooting
The following table lists common mapping problems and how to resolve them:
| Problem | Likely cause | What to do |
| :------------------------------------- | :-------------------------------------------------------- | :--------------------------------------------------------------------------------------------- |
| **Generate mappings** is unavailable | Endpoint mapping is still running | Wait until the banner clears |
| No property rows appear | The endpoint isn't approved, or schema mapping hasn't run | Approve the endpoint, then run schema mapping |
| **Mapping schemas…** never finishes | The operation is large, or the backend queue is busy | Refresh the page, then run schema mapping for that endpoint |
| The property list is empty after a run | The two schemas have no structural match | Check that you're on the correct **Request** or **Response** side, then map the fields by hand |
| A SOAP transformation is empty later | You approved only the response properties | Approve the request property mappings, which the XSLT generator uses |
| Actions are unavailable | Another workflow is running | Wait for it to finish. To see why an action is unavailable, point to it. |
## Next steps
* [Transformations and generated code](/connector-studio/transformations)
* [Design routes with Route Builder](/connector-studio/route-builder)
* [Connect from your IDE with MCP](/connector-studio/connect-with-mcp)