{"id":8426,"date":"2025-10-28T13:39:05","date_gmt":"2025-10-28T12:39:05","guid":{"rendered":"https:\/\/tacstone.nl\/?page_id=8426"},"modified":"2026-04-22T17:14:38","modified_gmt":"2026-04-22T15:14:38","slug":"ai-in-action-how-to-deploy-prompts-scalably","status":"publish","type":"praktijkvoorbeeld","link":"https:\/\/www.tacstone.nl\/en\/practical-examples\/ai-in-action-how-to-deploy-prompts-scalably\/","title":{"rendered":"AI in action: How to deploy prompts scalably"},"content":{"rendered":"[vc_row type=”full_width_content” full_screen_row_position=”middle” column_margin=”default” equal_height=”yes” content_placement=”middle” vertically_center_columns=”true” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” bg_image=”6442″ bg_position=”center center” background_image_loading=”default” bg_repeat=”no-repeat” scene_position=”center” top_padding=”0″ constrain_group_1=”yes” bottom_padding=”0″ text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” class=”pageHero” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/1″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in-from-bottom” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_row_inner content_placement=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” text_align=”left” class=”pageHero__content”][vc_column_inner column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/1″ tablet_width_inherit=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_column_text]\n
\n
\n

AI in action: How to deploy prompts scalably<\/h1>\n<\/div>\n<\/div>\n[\/vc_column_text][\/vc_column_inner][\/vc_row_inner][\/vc_column][\/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” top_padding=”60″ text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”2\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in-from-bottom” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_column_text el_class=”nooitklein”]\n

AI is smart, but only if you are.<\/h2>\n

Artificial Intelligence<\/a>(AI<\/a>) offers unprecedented opportunities for automating business processes. But the real power of AI only comes into its own when it is properly driven. The key? A good prompt. <\/strong> In this practical example, we show how a smart prompt makes a difference in reading sales orders, a task that previously required a lot of manual work.[\/vc_column_text][\/vc_column][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][\/vc_column][\/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” top_padding=”100″ bottom_padding=”100″ text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/1″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in-from-bottom” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_row_inner column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” text_align=”left”][vc_column_inner column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” el_class=”iconUSP__item” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in” border_type=”simple” column_border_width=”none” column_border_style=”solid” delay=”150″][image_with_animation image_url=”2403″ animation=”Fade In” hover_animation=”none” alignment=”” border_radius=”none” box_shadow=”none” image_loading=”default” max_width=”100%” max_width_mobile=”default” el_class=”image__icon”][vc_column_text]\n

Scalability without extra work<\/h3>\n

One well-formulated prompt can extract data from numerous different document formats without the need for templates.[\/vc_column_text][\/vc_column_inner][vc_column_inner column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” el_class=”iconUSP__item” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in” border_type=”simple” column_border_width=”none” column_border_style=”solid” delay=”300″][image_with_animation image_url=”2403″ animation=”Fade In” hover_animation=”none” alignment=”” border_radius=”none” box_shadow=”none” image_loading=”default” max_width=”100%” max_width_mobile=”default” el_class=”image__icon”][vc_column_text]\n

Efficiency and error reduction<\/h3>\n

Targeting AI minimizes errors and speeds up processes; from manual hours to automated minutes or even seconds.[\/vc_column_text][\/vc_column_inner][vc_column_inner column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” el_class=”iconUSP__item” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in” border_type=”simple” column_border_width=”none” column_border_style=”solid” delay=”300″][image_with_animation image_url=”2403″ animation=”Fade In” hover_animation=”none” alignment=”” border_radius=”none” box_shadow=”none” image_loading=”default” max_width=”100%” max_width_mobile=”default” el_class=”image__icon”][vc_column_text]\n

Seamless integration with existing systems<\/strong><\/h3>\n

AI output can be delivered directly in structured formats such as JSON, ready for processing in ERP systems via APIs.[\/vc_column_text][\/vc_column_inner][\/vc_row_inner][\/vc_column][\/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”2\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in-from-bottom” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_column_text]\n

Reason<\/span><\/h3>\n

Many organizations receive dozens to hundreds of sales orders<\/strong> via e-mail every day, often in PDF or text format. These orders vary widely in formatting and structure, depending on the customer or supplier. Previously, OCR technology combined with templates per supplier was used to extract the correct data. This process was not only time-consuming, but also error-prone and difficult to scale. Any change in a document structure required modification of the template, leading to delays and additional management burden.<\/p>\n

The rise of generative AI brought new possibilities<\/strong>. Can this technology be used to automate order reading without fixed formats?[\/vc_column_text][\/vc_column][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][\/vc_column][\/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” top_padding=”60″ text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”2\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” bg_image_animation=”none” enable_animation=”true” animation=”fade-in-from-bottom” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_column_text]\n

Approach<\/span><\/h3>\n

We started with an analysis of the existing order processing process and identified the bottlenecks<\/strong>: dependence on templates, limited flexibility and high error proneness. We then explored how AI models such as GPT could be used to overcome these constraints.<\/p>\n

The core of the approach lay in optimizing the prompt<\/strong>; the work instruction given to the AI. We experimented with different formulations to determine which prompts were most effective in extracting relevant information from unstructured documents. In doing so, we took into account four elements:<\/strong><\/p>\n