Web form validation hasn't change a lot in the last 20 years. Generally a form is set to check for required fields or in more advanced scenarios it is possible to make fields required based on conditions.

There are also ways to make sure that a field entry matches a set pattern. This is handy when you want to check for a valid email address or when you want to make sure phone numbers are entered using a consistent pattern. For many of us, these are all familiar scenarios in which we have plenty of pratice.

UltimateForms Form Designer provides excellent abilities to set up validation for all these scenarios. The new AI Answer function provides a new level of validation.

Imagine all the times that you have received form submissions with sparse details and incomplete information. It happens all the time. Conventional form validation isn't great at preventing this problem. The new AI Answer function adds a new layer of validation that can dynamically analyze form submissions to ensure correct and complete information in the first submission by analyzing user inputs before the form is saved.

Let's think of a typical scenario with a support request. Frequently users will submit a low effort request in which little information is provided describing the issue. This causes delays and frustration, because a human must ask for more information before they can begin researching the problem and working towards resolution. It is far better to eliminate low effort submissions by instead prompting the user for more information before the request is saved.

How can you begin to take advantage of AI Driven validation?

Begin with some practice by going through the examples in our previous article here: New AI Answer feature for Infowise Ultimate Forms SharePoint Forms (infowisesolutions.com)

AI is able to assess user entered content and return an evaluation according to your prompt. Let's focus on the description provided by the user. We're looking to stop low effort descriptions like "The program won't load". We want AI to evaluate the description to determine whether the user did a good job providing details. When doing this kind of validation, it is essential to provide a specific prompt describing the exact response required from AI.

We want the user to click a button under the description which will trigger the validation. You could put a button under the description with a label Confirm Description. The button should call the AI Answer function with the user data and return a "score" based on the user information. Here are some example prompts that you can try:

$Answer([Description]|Based on the description, evaluate the level of detail describing a problem on a scale of "1" being very broad and vague to "10" being very specific and detailed. Provide back only a single number response with values from "1" to "10".)

An essential part of the prompt is defining precisely what must be returned. Note that the returned data is text and must be written back to a text field in your form. You could call this field "AI Check", hide the label, make it read only, then set a conventional validation rule on this field to make sure it has an acceptable value. In our example, we might validate that the AI evaluation returns back 6,7,8,9 or 10 only. We can hide the save button blocking form submission until user clears the check.

In this way, we are letting AI screen the initial information cutting down on delays and wasted time where a human must gather this information later. This kind of validation needs some experimentation to see how it performs and to make sure that it will always return back an evaluation exactly with the responses specified. With some practice, you will become more comfortable with what works well in your prompts. Note that we could evaluate for anything that we need contained in the description such as make and model numbers, specific steps, the name of the software used, which browser was used or any other kinds of details you need to make sure are always given.

Here is an example of this new kind of AI driven validation being used in a support form: