Skip to main content

Get help for ARIS

Generate models with Companion

You can use text or images to generate models. This allows you to automatically transform process descriptions, interviews, or standard operating procedures into structured process models. The language selected during login or the language selected in the model header is used for generation, regardless of the language used in the prompt.

Reuse of database objects

AI-based model generation enables the reuse of predefined objects. This requires Companion to be enabled for the relevant database.

The object types that can be considered for reuse are satellites that can be assigned to a function or task. The list of supported satellites is specified in the content type configuration in the relation properties of the activity. If relation properties were changed, a reindex of the database must be performed under Publication settings (Administration Administration > Configuration Configuration).

Prerequisites

  • You have at least the Designer role.

  • You have the Companion user function privilege.

  • You have Write access privileges for the relevant database group.

  • The model must be of one of the following types: BPMN collaboration diagram (BPMN 2.0), BPMN process diagram (BPMN 2.0), Enterprise BPMN collaboration diagram, Enterprise BPMN process diagram, Event-driven process chain (EPC), EPC (row display).

Limits and formats

  • Prompt

    To measure the amount of text you entered as a prompt, your input is broken down into small data units called tokens. One token is the equivalent of approximately 0.75 words. The maximum number of tokens allowed for a prompt is 100,000 tokens, which corresponds to approximately 70,000 words.

  • Conversation

    The limit is 1,000,000 input tokens or 100,000 output tokens. Once either of these limits is reached, the conversation cannot be continued.

  • File upload

    • Maximum page limit of 50 pages and data volume limit of 10 MB. Large images or PDF pages are split into several smaller tiles or pages. This means that even documents with fewer than 50 pages can reach the 50-page limit.

    • Allowed file formats:

      DOCX, DOC, VSDX, VSD, PPTX, PPT, PDF, TXT, JSON, XML, CSV, BPMN and XLS/XLSX (not supported: Excel files with complex formulas or errors.

    • Allowed image formats:

      PNG, JPG, JPEG, and BMP.

Procedure. Procedure

    Step 1 - Generate a model

    1. Open a model or create a new model.

    2. Click Generate from text Generate process. Companion Companion opens.

    3. Enter or paste the text describing the model you want to create or the changes to an existing model into the prompt box.

      Alternatively, drag one or more files into the prompt box or click Upload file Upload file. Multiple input files are processed as a single file, which means that information from various files is combined, with overlapping information resolved accordingly.

    4. If available, provide the modeling conventions that are to be complied with.

    5. Select the AI reasoning mode. The Thinking mode requires more time than the Fast mode, but the quality of the result is usually better.

    6. To automatically add descriptions for the process or the objects, select Add description for process and/or Add description for functions/tasks.

    7. Click Generate model.

      The model is displayed as a preview. Reused database objects are highlighted with a colored frame.

    Step 2 - Review result and make changes

    1. Check the result and decide whether any changes are required. For example, you can

      1. restructure the process flow

      2. revert the reuse of database elements

      3. rename elements

      4. add, remove, replace, or update activities/events, satellites, branches, or conditions.

    2. Type what you want to modify into the input field. The model is updated based on your input, and a summary of the changes made is displayed.

    3. Use the object modifications listed in the summary, to easily identify and review what was changed. For example, click an object in the summary, to highlight it in the model preview. This is possible for the following objects: functions, tasks, gateways, and events, as well as satellites like roles, IT systems, risks, documents, and so on.

    4. Companion proposes further possible modifications, if there are any. If appropriate, click one or more of the suggestions to transfer the suggestions to the input box, and then press Enter. Alternatively, ignore the suggestions and enter your change request directly to the prompt box.

    5. Continue to ask for adjustments and improvements until the result meets your requirements.

      The model is displayed as a preview.

    Step 3 - Finalize the model

    1. Use Undo/Retry Undo and Redo Redo to go back or forward up to ten steps in the conversation with Companion. When you undo a change, the associated conversation is also reset as if you had never made the request. The same principle applies when you redo a change.

    2. If you are content with the result preview, click Insert into model. The credits for the model generation will be consumed.

    3. If the Description/Definition attribute of the process was already maintained, you can decide whether to overwrite it or to keep it. Your generated content is inserted in the model. If the model already contains objects, the new content is placed next to or below them. You can now adjust the generated content manually or connect it to existing objects.

    4. Click Save.

    Your generated content is inserted and saved as a model.

    As a Companion admin you can check the remaining credits.