Skip to main content
Roberty Studio
IDPIDP

Playground and Simulate API

Test an IDP template's draft before publishing: the Playground, which shows each stage's result, and Simulate API, which builds the call the way an external system would.

Playground and Simulate API

A template offers two ways to test the draft — the builder's current configuration, with changes not yet published — on a test document. Both sit in the Test menu at the top of the template, and at the end of the builder, in the Result stage.

ToolWhat it is for
PlaygroundRuns the draft on a sample document and shows each stage's result.
Simulate APIBuilds the call the way an external system would and shows a summary of the processing.

Both run the draft, not the active version, and record the run in the Development environment, under Runs. To test several documents at once and compare them with the expected outcome, use the evaluation sets.

Only workspace developers and administrators can run tests.

warning

A test run is a real pipeline run: it counts toward the workspace's IDP request quota, uses OCR and AI according to the stages, and robot validations really run the robot.

Playground​

Inputs​

InputDescription
Test documentA PDF, PNG or JPG, dropped onto the Playground or picked from your computer. The template's reference document is not loaded automatically, and the test document is not saved on the template. Use Change to test another file.
PageThe pages of the test document to process, with the same syntax as the analysis scope, for example 1-3, 5. Empty processes the pages in the template's scope. Only shown when the Allow the API to override pages per run option is on; see Document and pages.
Parameters (JSON)Optional. A JSON object sent along with the document, like the parameters of an API call, for example {"customer": "ACME"}. It must be an object; any other value is refused.

Click Run to run the pipeline. The Playground follows the run to the end and gives up after 5 minutes without an answer, with a timeout message; the run is still recorded under Runs.

Result per stage​

When it finishes, the Playground shows one tab per pipeline stage. Stages that did nothing with the document show as empty on the run's map.

TabWhat it shows
DocumentThe test document you uploaded.
Pre-processingThe image after each filter and the final image.
IdentificationThe similarity to the reference, whether the document was accepted or rejected, the detected type and fraud signals. When the document is rejected, it shows Pipeline stopped, and the following stages show as not run.
ExtractionThe Extracted text, Forms, Tables and Text blocks, depending on the method, with a search over the content. Blocks show their confidence — high from 90%, medium from 70% — and position; the page image shows page 1 only.
Post-processingEach step's output, in order.
ValidationsThe score, for example "2/3 passed", and each validation's result. For robot validations, the arguments sent and the robot's reply.
ResultThe pipeline's final output, formatted (Pretty) or raw (Raw), and the run's total time.

The Playground also feeds the editor: after a run, the post-processing and validation code editors and the robot input mapping suggest the fields of the result and of the parameters.

Simulate API​

Simulate API opens the Simulate processing window, which builds the call the way an external system would: a file, the pages (when the page override option is on) and the Parameters, an optional JSON object.

It accepts pdf, png, jpg, jpeg, webp, tiff, tif and bmp files. Click Run pipeline to process it.

The result shows a summary — Total time, Identification (accepted or rejected), extraction Method and passed Validations — and one tab per stage: Identification, Pre-processing, Extraction, Post-processing and Validations.

Simulate API runs the current draft, in the Development environment, not the active version. When the template has no published, active version yet, the window shows a notice, but the simulation runs all the same. To call the template through the real API, see API and webhook.

When to use each​

UseWhen
PlaygroundYou are building or adjusting the pipeline and want to see what each stage did: the filters, the extracted text, the output of each post-processing step.
Simulate APIYou want to check the result the way an integration would see it, with the file, pages and parameters the external system will send.
Evaluation setsYou want to make sure a change did not break documents that already worked, before publishing.