The sample scenario and demo in this material were prepared using our sandbox environment, Kbank. If you'd like to follow along, you can request the demo website here (opens in a new tab), choosing the Xperience by Kentico - Business Tutorial Kbank demo site option.
AIRA includes several AI-assisted features in Xperience by Kentico. This material focuses specifically on website content generation and refinements, which help you create and improve website copy directly in supported text fields.
For business users, the main value is speed and efficiency. Instead of starting from a blank field every time, you can use AIRA to generate a draft from the context already available in your content item, then refine that draft until it is clearer, shorter, stronger, and better aligned with your message.
This feature works best as a practical writing assistant inside your existing editing flow. It can help you move faster, but it does not replace editorial judgment. You still need to review the final copy for accuracy, tone, and fit before publishing.
What this guide helps you do
Use AIRA to create and improve website copy directly in your editing fields, faster and with better consistency.
Use content generation when you need a fresh starting point. This is especially useful when a field is empty, and you already have helpful context in the current item.
Use refinement when the field already contains text, and you want to improve it. This is the better choice when the message is mostly right, but needs editing.
A simple way to think about it is:
In many real workflows, you will use both. Generate a starting point first, then refine it until it feels ready for review.
Before you begin, make sure the feature is available and that the content item includes useful context.
AIRA can use several inputs when generating content for a field:
Better context produces better drafts. Include information like page goal, target audience, offer summary, key value, and CTA intent.
If the field that contains context is left blank, generation from the current item into your target field will not work. In that case, you can still use a custom prompt in AIRA chat window.
Want to see how content generation and refinements work in practice first? Then watch or follow along with a short demo that showcases a simple flow:
If you're ready for the full scoop, continue reading below.
Imagine you are creating copy for a landing page section that promotes a marketing webinar.
Use this sample scenario:
This scenario works well because it represents a common business task, like turning a page goal into clear, action-oriented website messaging.
AIRA performs best when the content item already contains key information.
Before you generate website copy, make sure the surrounding context gives AIRA enough direction, and prepare or verify these materials in the item:
You do not need to overcomplicate the setup. A few clear and relevant inputs are usually more useful than a large amount of vague background text.
Here is an example of source information that can lead to a stronger first draft:
If your first draft feels generic, the problem may be a lack of or weak context, not the generation step itself.
When a supported field is empty, you can use the Generate content action, marked with the AIRA button. There are two main ways to generate content from an empty field.
Use this option when the content item already contains relevant context.
This option is usually the fastest route when your item already contains enough useful information.
Use this option when you want more control, or when the current item does not contain enough context.
This option gives you more flexibility, but it is a bit more manual because you need to copy the output into the field yourself.
Use item-context generation for speed. Use a custom prompt when you need to guide the output more precisely.
At this stage, do not expect the final copy. Focus on whether the draft is useful and can be efficiently refined by AIRA, or by human editing.
Ask yourself:
If the answer is mostly yes, move to refinement. If the answer is no, improve the context of the current content item, or try another draft.
When a field already contains text, you can use the Refine text action, marked with the AIRA button.
One of the available actions is Regenerate content.
The Regenerate content action completely rewrites the existing text in the field. If you want to keep an earlier version, copy it to a safe location before regenerating the content with AIRA.
Use Regenerate content when:
A good rule is this:
Because regeneration uses the available context from the item, you may get a better result if you improve the supporting context first.
When the field already contains text, you can use the AIRA button to open Refine text and choose from available refinement options.
Apart from regenerating the field's content, the refinement options can help you:
These refinements apply to the entire field.
A simple refinement flow for website content could look like this:
This keeps the process efficient and prevents unnecessary rewriting.
If rich text editor refinements are enabled, you will also be able to refine only a selected part of the text instead of the whole field.
This is especially useful when most of the content is already good, and you only want to improve one paragraph, sentence, or a few words.
The basic flow is:
This can be helpful in Page Builder and other editing views that use rich text fields.
Whole-field refinements replace the full field content. Rich text editor refinements can help when you want a smaller, more targeted edit.
Preset refinement actions are useful, but sometimes you need more control. In those cases, use Apply custom prompt.
With this option, you describe the desired change in the AIRA chat and then paste the result back into the field.
Custom prompts are especially helpful when you want to:
The most effective prompts are direct and specific. Tell AIRA what to keep, what to change, and what outcome you want.
The custom prompt workflow is different from in-place refinements. You review the result in the AIRA chat and then copy it back into the field manually.
To get you quickly on track, use the prompts below as a starting point in your project. Replace the bracketed placeholders with your own content, and use as is, or update the prompts as much as needed. Don't be afraid to iterate on prompts that work for you to get the best results.
Write website copy for [audience]. Goal: [goal]. Key value: [value proposition]. CTA: [cta]. Tone: [tone]. Keep plain language and avoid buzzwords.
Use this when you want to shape the direction of a first draft.
Generate 8 landing page headline options for [audience] promoting [offer]. Focus on clarity and value. Keep each under [X] characters.
Use this prompt when you want to quickly compare several headline alternatives.
Draft a short landing page intro for [audience] about [offer]. Include one key benefit and one clear CTA. Keep it concise and skimmable.
Use this for event pages, webinar pages, or feature highlights.
Keep the meaning, but reduce this by about 30%. Use shorter sentences and preserve the CTA intent.
Use this when the message is correct but too long or too dense, and you want to control how much shorter you need it to be.
Rewrite for clearer, simpler language at a broad business reading level. Keep the same message and tone.
Use this when the copy sounds abstract, formal, or hard to scan.
Keep the message the same, but make the CTA more specific and action-oriented. Avoid sounding pushy.
Use this when the text needs more momentum without becoming too aggressive.
Keep the meaning, but make the wording more confident, practical, and professional. Remove vague claims and unnecessary hype.
Use this when the draft feels too generic or too promotional.
Let's return to our demo scenario: a landing page with an invitation to a webinar. The example below shows how a rough message can become clearer and more useful through refinement.
Headline:
“Transform your organizational operational marketing outcomes with strategic automation.”
Intro:
“Our upcoming knowledge session provides an opportunity to gain numerous insights that can potentially improve several business outcomes in your organization.”
CTA:
“Learn more”
Headline:
“Improve content results with smarter automation.”
Intro:
“Join our webinar to learn practical ways to improve content performance with less manual effort.”
CTA:
“Reserve your seat”
Regeneration and in-place refinements change the content in the field, but the changes are permanent only after you publish the item. That means your final workflow before publishing should always include these last steps:
This is a good moment to make sure nothing important was lost during generation or refinement.
The system keeps changes only after you save the item – including changes you manually make to an existing draft. Regeneration and refinements overwrite the existing field content.
If you worked on a draft, saved your manual changes before you made any AIRA-powered refinements, and now want to return the item to the version before AIRA refinements, you can still do so by using the Revert to published option.
However, if you make manual changes on an existing draft and do not save them before refinement or regeneration, the AIRA-powered refinements will fully overwrite your manual changes, and you will not be able to restore them.
Even when the output looks great, you should still perform a final human review before publishing. This is an important part of the workflow, not an optional extra step. As with any AI-assisted workflow, human signoff improves the accuracy of the content as well it's natural flow.
Before you publish, check the final copy for:
Length limits guide AIRA during generation, but your review and final edits ensure the content is publication‑ready.
Once you get to try the sample workflow, you can apply it to other daily tasks related to website content creation.
Examples include:
These scenarios use the same pattern: prepare context, let AIRA generate a draft and refine the result, and then review the final copy before you publish it.
Generated copy can speed up drafting, but some content types need extra care. Do not publish the generated website text without closer review when it includes:
In these cases, use AIRA to accelerate the first draft or improve wording, but keep human review and approval firmly in the process.
AIRA-generated output should always be treated as a draft. Final responsibility for accuracy, clarity, and brand fit stays with your team.
AIRA saves time during first drafts and refinements of existing content. But it works best when you understand what affects the output quality and how you can help improve it.
If the surrounding item does not contain useful information, the generated text may feel generic or incomplete.
To improve results:
Generated copy may be longer than ideal for the field or page layout.
To improve results:
Sometimes you need one or two additional iterations before the text feels right. Don't be afraid to experiment to find an approach that works best for your project and your audience.
To improve results:
AIRA interactions are commercially protected. Customer data inputs, prompts, and generated responses are discarded after each interaction and are not used to train the underlying models.
That being said, you should still follow any internal rules about sensitive content and review your content carefully before publishing.
Use this list of final checks before you publish website copy that was created or refined with AIRA.
AIRA can help you move from blank fields to usable website copy much faster, especially when you already know your audience, goal, and core message. The strongest results usually come from a simple workflow:
For marketers and content editors, AIRA helps reduce manual effort, improve consistency, and deliver stronger website copy faster, while preserving space for human judgment and final decisions.
If you use AIRA this way, content generation and refinements becomes an organic part of your workflow for landing pages, webinar pages, product messaging, and other everyday website tasks.
Eager to get more out of AIRA's AI-powered capabilities?