AI Process Generator FAQ

Note: AI Process Generator is currently in beta testing and is available on all tiers, but is not available in the AEN region. It will only be available to the Professional, Expert, and Enterprise tiers on full release.

The following questions and answers deal with general functionality of this generative AI feature:

You now can choose to create your procedure manually or by using generative AI. Using the generative AI option relies on a text-based prompt which is synthesized from the process summary information and any additional instructions you specify. This generates a suggested first draft of the process, including activities, tasks, and role assignments. You can then choose to discard the outcome, regenerate the procedure, modify your prompt, or accept the procedure as the first draft of your process.

The prompt uses the following information from the process summary:

  • Process Name

  • Process Objective

  • Process Background (if available)

  • Roles of Process Expert and Owner

  • Process Group Name

Additionally, before generating the procedure, you can provide additional instructions that will only be used for the procedure generation and will not be saved.

The prompt is composed of information from the process summary along with additional information that you provide when using the option. To modify your process summary information, you must go to the Process Summary tab, edit the information and save a new draft. They can then navigate back to the Procedure tab to re-launch the Start with AI feature.

If the generated procedure does not meet your expected outcome, you have the following options:

  • Regenerate the procedure: Due to the nature of generative AI, simply regenerating might provide with a different outcome.

  • Modify the prompt: You can go back and provide additional instructions, or change the information in the process summary (see the question above for steps to do this). Then regenerate the procedure with the new information.

  • Edit manually: You can accept the draft and then edit it manually in the usual process editing environment.

No, this functionality only applies to new procedures that have no activities added to them.

By utilizing the power of OpenAI, this feature is designed to be versatile and can draft a wide range of business procedures, from standard operating procedures to business process workflows. The quality of the output will depend heavily on the quality of the information inputted in the process summary and in the additional instructions provided.

We are always happy to receive your feedback. You can raise any new functionality ideas through the Nintex Ideas portal, contact your Account Manager, or raise any issues with support@nintex.com.

Like any feature utilizing AI, this feature has limitations. It may not fully understand complex or ambiguous requirements, and the quality of the draft generated depends on the quality of the input provided. Additionally, the prompt will always consider all the information provided in the process summary by default and there is no prompt history tracking.

Yes. Please see the character types in the table below:

Invalid Character type Examples
Symbols $ @ # & * % + = / \
Currency ¥ € £ ₹ ₩ ؋ ₴ ₱
Math Operators × ÷ ∞ ≈ ≠ ≤ ≥ ± ∑ π
Emojis ☺❤(any emoji)
Control Characters \t \n \r \0 \u200B \uFEFF (tabs, newlines, zero-width space)
Formatting @ ⑩ ① ② ③ ✖
Miscellaneous ♂ ♀ ♪ § ¶ ‡

No. Currently this feature is only used to generate procedures.

These questions and answers deal with how data is handled while using the feature:

The information from process summary and any additional instructions are synthesized in a prompt, which is then passed to a private MS OpenAI instance hosted on the Nintex Azure Infrastructure. Using the prompt this service generates a suggested procedure, which is returned to Process Manager, where it will be interpreted and displayed for you to review.

No, all the data is kept inside of Nintex controlled spaces and the same Process Manager trust boundaries will apply to this feature.

No prompt or process specific data is collected by Nintex. The only data collected as part of this feature is basic usage data to measure the quality of the interaction (such as has the user accepted or regenerated their draft).

To deliver our AI Process Generator functionality we use the Azure OpenAI Service, using the GPT-4o mini model. We use Azure OpenAI for the security capabilities of Microsoft Azure, while running the same models as OpenAI, within our Nintex MS Azure hosted instance. All data remains within the sameProcess Manager trust boundaries that apply today.

The knowledge bank used is the one used to originally train the OpenAI GPT-4o mini model. See the GPT-4o System Card for more information. No additional knowledge bank or real-time data are used.

No prompt or process specific data is being used to train the Large Language Model (LLM) used to enable this functionality.

These questions and answers deal with the availability of the feature:

Check the feature availability based on the tier package of your Process Manager license. This feature is not currently available in the AEN region (United Arab Emirates).

This functionality is only available to users with the right to create new processes (i.e. Process Editors).

There is no explicit limit of using the generative function. The only limit is process entitlement, as each saved process will count towards your consumption limit.