AI Conception

• The conception of the construction of machines, websites, apps or even AI is the most important organizational task.

• We will guide you through it in 5 steps.

• First, you need to know that every AI response follows a structure that can be influenced.

• Finally, you will be provided with a Template to start your conception.

1. Each AI response is broken down as follows

1. Header

Introduction or introduction, e.g. greeting, situational context or hook

2. Body

Main content with product information, answers, arguments

3. Footer

Conclusion with, for example, CTA, links, further recommendations, imprint, etc.

The following image shows you the header, body and footer from top to bottom.

2. What does the maturity model do for you?

The maturity model describes 7 levels of how precise and context-sensitive your AI answers are. Each stage expands quality with new dependencies – from simple product type to dynamic, campaign-based communication.

We recommend starting with a maximum of the first two levels.

3. In which cases should the AI response be even more intelligent or more tailored to the query?

Product dependency

• Do you want the structure of the answer to vary depending on the product type?

• Is there any special information, approvals or standards that must appear in the header, body or footer for certain products?

• Should certain technical features (processor, performance, graphics card) always be highlighted?

Example

• Header: Highest battery power up to 48h – our travel laptop for extremely long journeys"

• Body: Table with column X, column Y, column..."

• Footer: Download Safety Data Sheet: [PDF link]

Keyword Triggers

• For example: Depending on the keyword such as "light", "high performance", "waterproof", "contact", certain response templates trigger

• Are there keywords that need to trigger legally relevant notices?

• Are there keywords that should trigger a comparison?

Example

• Header: Thank you to get contact and get more information about the product Travel Laptop sign up as below:

• Body: First Name Last Name | Position | Phone number | Email

• Footer: Alternative like to write to the support@foxbase.de

4. Data Enrichment | Even the AI can only answer what it knows!

In order to enable maturity levels, existing data will probably have to be enriched . You can find out how to enrich data in the following article.

Now an example

• I want to have a product dependency where the footer contains the correct contact details.

• I want to have a keyword dependency where the word "contact" shows a detailed contact card in the body.

  The table shows your existing data - even your PDF documents are always converted into a table in the background.

  If your documents don't contain any contact details, it will be difficult for the AI to intelligently answer which product area which contact person should be displayed. If there is no phone number, no phone number will be displayed.

In order to display product-specific or keyword-specific contact details, contact details must be added to the data. To do this, you can either use AI or manually enrich your data.

• AI-supported: You explain where in the PDF documents the contact details for the respective product are located. For example, always at the bottom right of the company logo.

• Manual: You create an Excel that contains the product names in column A and the contact details in another column.

• As a reminder, if there is no phone number, then the AI will never be able to output the correct phone number as an answer.

• Check out the related article to enrich data effectively.

5. Conclusion and template work

• To move forward, you can download the template to help.

• Put yourself in the position of if-then logic. If product category A is answered, then the following information should be displayed in the header.

• Download Header Body Footer Requirements

Once these requirements are completed, based on them:

✅ Missing data and information can be enriched

✅ A prompt kit can be created

✅ Individual AI response templates are developed

✅ Effective testing is made possible. Is AI clever enough without the Prompt Toolkit?

✅ Coordinate future automation logic (e.g. in CMS, various chatbots or PIM system)

More examples of AI response quality in the maturity model