dAI Pro – images, diagrams and documents from your knowledge

The editorial copilot of dAI Pro already writes texts from the knowledge of your collection. Now images and finished documents are created there too – an illustration in your brand style, a diagram with exact labels and figures, or a document as PDF and Word on your letterhead. Every result rests on the contents of your collection, names its sources, marks missing knowledge and carries the label as AI content. New are variants, editing with a hint and areas as well as text as a separate layer with editable SVG and PowerPoint files. You find all functions on the dAI Pro features page and an introduction with use cases on the dAI Pro product page.

You describe in the copilot what you need – for example “troubleshooting when the heating does not get warm”. dAI Pro finds the relevant passages in your collection, has the language model write a brief from them and implements it in your corporate design. You see the result as a preview before anything is published.

• Three task types – illustration, diagram and document as PDF and Word. In addition, the “As PDF and Word” button turns any finished text draft into a document.
• Knowledge instead of invention – labels, figures and factual statements come from your content. Anything missing appears as a gap in the result, not as a made-up detail.
• Any format – fourteen standard formats from square to LinkedIn banner, a custom size in pixels and a resolution from 1× to 4×.
• Files for the job – a product logo, a sketch as starting image or a different letterhead apply to this one job only.
• You decide – keep it, regenerate with a hint or discard it. Drafts that are not kept are cleaned up by dAI Pro after one day.

For headers, posts and presentations, dAI Pro creates images with an image model of your choice – gpt-image-2.5 from OpenAI or Google Gemini, assigned per collection. The language model writes the image brief from the briefing and the knowledge base, and dAI Pro adds colours, background and rules from your corporate design.

• Brand colours and style – palette, image style and a list of things to avoid go into every brief. Reference images additionally show the model your style.
• Exact background – with OpenAI the subject is created as a cut-out. dAI Pro fills the background pixel-exact in your colour or delivers a transparent image.
• Your logo in the image – placed unchanged at a chosen position or built into the scene by the image model, for instance on a screen or a sign.
• Starting image and style template – an existing image can be redesigned, and a style template applies to this job only.
• Self-check – dAI Pro checks the finished image against the brief – labels, number of elements, background and logo. If it deviates, it is generated again exactly once.
• Alt text and caption – come along right away, in the language of the job.

More control over the result comes from variants, editing with a hint and areas as well as text as a separate layer.

With OpenAI’s gpt-image-2.5 models, dAI Pro gets more out of every job. You watch the image take shape, order several drafts at once and change a good subject deliberately instead of rolling the dice again. The interface only shows what the assigned image model can actually do.

• Five quality levels – from draft (fast) to maximum. The fast level is enough for a first take, the high one for print.
• Variants – two, three or four drafts in one job, each with its own self-check. You pick the best, the others disappear after one day.
• Intermediate images – while the model draws, dAI Pro shows the image in stages. You see early whether the subject is heading the right way.
• Edit with a hint – “make the needle red” or “replace the word Begin with Start”. The image model receives the finished image and changes only what you named. Alt text and caption follow, and Before and Edited sit side by side for comparison.
• One area only – drag a rectangle over the preview and only that section changes. Everything outside stays pixel for pixel.
• Declined jobs – if the provider declines a subject for content reasons, dAI Pro says so with stage and category instead of a technical error. No image is charged.

Google Gemini offers neither variants nor areas. There it stays with a new run plus hint, and the interface hides the other buttons.

Image models set type unreliably – letters slip, words come out wrong, and the typeface is never your corporate font. dAI Pro therefore separates image and text. The image model draws a subject without any text, and dAI Pro sets the captions afterwards in the fonts of your corporate design.

• Planned, not painted – the language model defines title, subtitle and captions with their position. The subject gets calm areas in those places.
• Your fonts, your colours – the layer uses the brand fonts you uploaded and checks the contrast against the background.
• Editable as SVG and PowerPoint – besides WebP and PNG you receive an SVG file with real text elements and a PowerPoint file with the image as background and one text box per caption. Change the texts in Illustrator, Inkscape, PowerPoint or Keynote without regenerating the image.
• Editing stays possible – a hint changes the base image without text, then dAI Pro sets the layer again. A changed title ends up in the layer, not in the image.
• Complete in the file list – when you keep the image in TYPO3, SVG and PowerPoint come along.
• Checked and labelled – the self-check inspects the text-free base image, and every version carries the AI label.

Where labels, order and figures have to be right, no image model draws. The language model supplies only the structure, and a fixed template designs it in your colours, with your fonts and your logo. That is why the diagram shows exactly what your collection provides.

• Six templates – process flow, step-by-step guide, comparison table, timeline, hierarchy and key figures.
• Flows with yes/no decisions – diamonds with side paths, jumps back and their own end, typical for troubleshooting and operating instructions.
• Backed entries – every station refers to its source in your collection. Anything missing is listed as a gap.
• More logos in the header – next to the logo from your corporate design, which you can also switch off for any job.
• Starting image – a sketch or an old diagram sets the structure.
• Sharp in any size – the render service draws the diagram in exactly the chosen format, also at 4× resolution for print.

dAI Pro sets guides, data sheets, FAQs, fact sheets and reports directly as a document. It is created twice – as a PDF and as a Word file, in your corporate design and on your collection’s letterhead.

• Letterhead recognised – dAI Pro finds the free area on the uploaded letterhead by itself, separately for the first and the following pages. Margins and page numbers can be adjusted.
• Your Word template – if a .dotx template is available, the Word file is created in it – with your styles, headers and footers.
• Diagrams and images in the text – dAI Pro inserts suitable diagrams by itself, and existing illustrations and uploaded logos where they fit.
• Citations and list of sources – factual statements carry superscript numbers, and the sources are listed at the end. Gaps are highlighted in the PDF.
• Template for this document only – a different letterhead or a different Word template, for instance for a subsidiary.
• From any text draft – the “As PDF and Word” button turns a finished copilot text into a document without a new model call.

Every image and every diagram is created in the format you need right now.

• Fourteen standard formats – square, landscape 16:9, 4:3 and 3:2, portrait 3:4, 2:3 and 4:5 for social media, story 9:16, link preview image, ultra-wide 21:9, banners 3:1 and 4:1 as well as A4 portrait and landscape.
• Custom size – width and height in pixels, from 64 to 4,096 pixels per side.
• Resolution 1× to 4× – up to 8,192 pixels per side. A4 at 3× resolution has about 300 dpi for print.
• Cropped to fit – gpt-image-2.5 delivers free sizes up to 3,840 pixels per side, older models only fixed formats. dAI Pro orders the closest size, crops it centrally and says openly when an image had to be enlarged.

You store colours, fonts, logo, reference images, letterhead and Word template once per collection. Illustrations, diagrams and documents use them automatically. Fonts are embedded, so the result matches your brand even without an internet connection.

• Files checked safely – SVG files lose scripts and external references, and Word templates with macros are rejected. The files are never publicly accessible.
• Check the letterhead – a sample page shows in advance where the text will sit on your letterhead.
• Files per job – up to four images and one template for exactly one job, deleted again after one day.

• AI labelling under Art. 50 of the AI Act – every image, every PDF and every Word file carries the machine-readable label. dAI Pro reads it back before delivery and releases nothing without it.
• Keep in the file list – in TYPO3, images and documents land in the file list with alt text and description and can be inserted into any content element. With text as a separate layer, SVG and PowerPoint come along.
• Costs in view – every job appears in the collection’s usage log, and images are billed by the provider’s tokens. Higher quality levels and variants cost accordingly more, and a monthly limit caps the number of illustrations.
• Provider of your choice – you choose the image model per collection, for instance gpt-image-2.5 flare or sunburst from OpenAI or a Google Gemini model. Diagrams and documents need no image model at all, only the collection’s language model.
• You have the final say – nothing is published automatically.

Illustrations, diagrams and documents are part of dAI Pro – in the TYPO3 extension as well as in the web application. The editorial copilot chapter of the manual explains how to set up and use them. With the demo with your own website you create a collection in a few minutes, and in a first conversation we will gladly show you the image features on your own content.