Chapter 24 covers two specialized AIUNIFY PROS SUITE Core Tools:
Although both use AI to transform existing content, they serve very different purposes.
Translation Hub converts text—or text detected from an image—into another language.
Website Wiki takes content extracted from a supplied website URL, restructures it with AI, and creates an editable Wiki-style document that can be exported as Markdown, PDF, or DOCX.
The Sidebar identifies this Tool as:
Translation Hub
The workspace itself is titled:
AI Translation Tool
with the description:
“Translate text between multiple languages with high accuracy and natural language processing.”
Navigate to:
Sidebar → Core Tools → Translation Hub
The route is:
/translator
The current page requires an authenticated PROS SUITE session. If you are not signed in, it redirects toward /login.
The current /translator page verifies authentication but does not contain its own visible FeatureGuard wrapper in the supplied source.
This differs from some other PROS SUITE Tools such as Game Developer.
Actual visibility and availability can still depend on current system configuration and Sidebar access rules.
Translation Hub is not presented through the Standard AI Generator interface from Chapter 12.
Instead, it provides dedicated controls for:
The current Translation Hub interface contains:
15 languages.
These are the current user-selectable language options verified in the workspace.
The current list is:
A generic internal Tool description elsewhere in PROS SUITE says the Translator can translate between 50+ languages.
However, the actual current Translation Hub workspace exposes only the 15 languages listed above.
For this User Operations Manual:
The current visible 15-language selector is the authoritative subscriber-facing workflow.
Do not assume an unlisted language can be selected through the current interface.
The first major Card is:
Language Selection.
It contains:
The source-language selector is labeled:
From (Optional).
You do not have to identify the original language manually.
When no source language is selected, the field displays:
Auto-detect.
In that case, Translation Hub instructs the AI to treat the source language as automatically detected.
Auto-detect is useful when:
Select a specific source language when:
The destination field is labeled:
To *.
The asterisk reflects that a target language is required before Translation can begin.
If no target language is selected, Translation Hub displays:
Error
“Please select a target language.”
Between the two language selectors is the Swap control.
It reverses:
when both have been explicitly selected.
The Swap control is disabled unless both:
contain selected values.
Therefore, you cannot use Swap while the source field remains on Auto-detect.
Swap does more than reverse the language selectors.
It also exchanges:
Suppose you translated:
English → Spanish
Original:
Good morning. How can we help you?
Translation:
Buenos días. ¿Cómo podemos ayudarle?
After Swap:
This makes reverse translation convenient.
Swapping fields does not immediately launch a new AI translation.
After reviewing the reversed configuration, select:
Translate
again when you want another AI-generated result.
The left input Card is titled:
Original Text
with:
“Enter the text or upload an image you want to translate.”
The Textarea placeholder is:
Enter text to translate...
The current Translation Textarea does not establish:
Very large translation requests can still be affected by the selected AI Model's practical context limits, so break unusually long documents into appropriate sections when needed.
Original Text also provides:
Add Image.
This allows Translation Hub to translate text found inside an image.
The file picker accepts:
image/*
The verified Translation attachment type is therefore an image.
The current maximum image size is:
5 MB.
If the image exceeds 5 MB, Translation Hub displays:
File Too Large
“Please upload an image smaller than 5MB.”
After an image is selected, a thumbnail appears inside the Original Text area.
Select the X on the image thumbnail to remove the pending image before Translation.
When Original Text contains written text, Translation Hub builds an instruction that asks the AI to:
If the Original Text field is empty but an image is attached, the instruction changes to:
Translate any text found in the uploaded image to the selected target language.
The AI is again instructed to preserve meaning and tone.
Translation Hub can therefore operate with:
The Translate button becomes available when one of those inputs exists and a target language is selected.
The current interface also allows text and an image to exist at the same time.
When written source text exists, the primary generated instruction focuses on translating that written text, while the image is still included with the AI request.
For predictable image-text extraction, use image-only mode when the main objective is specifically to translate the words visible inside the image.
The source does not expose a separate subscriber-facing OCR engine or OCR settings panel inside Translation Hub.
The image is supplied to the selected AI-generation workflow.
Therefore, image-reading quality can depend on:
For better results:
Translation Hub provides three:
Quick examples:
“Hello, how are you today?”
“Thank you for your help and support.”
“I would like to schedule a meeting for next week.”
The three exact examples are defined in the current Translation workspace.
Selecting an example places it in:
Original Text.
It does not automatically choose languages or start Translation.
Once:
select:
Translate.
While the AI is processing the request, the button changes to:
Translating...
The button remains disabled while the stream is active.
If Translation is attempted with neither text nor an image, the current Error is:
Error
“Please enter text or upload an image to translate.”
The AI is instructed:
“You are a professional translator. Provide accurate, natural-sounding translations that preserve the original meaning and tone.”
Translation Hub uses the shared AI streaming workflow.
As translated text arrives, it can appear progressively in the Translation output rather than requiring the entire response to finish first.
The right-side Card is titled:
Translation
with:
“Your translated text will appear here.”
Before a Translation exists, the output area displays:
Your translation will appear here.
Translation output contains two tabs:
Preview displays the generated Translation through the PROS SUITE Markdown Renderer.
Select:
Edit
to work directly with the translated text.
Once AI streaming has finished, the Edit field becomes directly editable.
You can correct:
before copying the final result.
While the AI stream is still running, the output Textarea is:
read-only.
Wait until Translation finishes before manually editing it.
After Translation completes, the Edit view provides:
Copy Translation.
When successful, the current notification is:
Copied
“Translation copied to clipboard.”
If you manually correct the Translation before selecting Copy, the current edited translatedText value is copied.
This allows the AI-generated draft and your final reviewed wording to remain part of one workflow.
The current Translation Hub does not provide a dedicated:
control.
The verified output-preservation action is:
Copy Translation.
There is no source-verified:
Save Translation
button.
The current workspace does not establish:
Translation state exists in the current page session.
Before leaving or refreshing the page:
Translation Hub uses the common useGeminiStream system, which reads the subscriber's currently selected PROS SUITE Model.
The shared streaming workflow also attaches configured provider API Keys where present.
Therefore, Chapters 10 and 11 apply to Translation Hub when the selected Model requires a personal provider credential.
Translation Hub does not have its own Model menu.
Use the normal Header Model Selector.
If you are translating text contained in an image, select a Model that supports the applicable visual/image input workflow.
The source verifies that the image is sent to the shared AI stream but does not guarantee identical image support across every selectable provider Model.
The general PROS SUITE feature catalogue lists:
Translation Hub — 10 Tokens.
The specialized Translation Hub calls the shared generation function without supplying a Translator-specific Tool ID.
The shared request therefore sends whatever Tool ID was supplied—in this case none from the Translation page.
The final server-side Token-accounting implementation is outside this User Operations bundle.
Therefore:
The catalogue's 10-Token value should not be presented as proof that every current Translation request is always billed exactly 10 Tokens.
The common streaming workflow refreshes the authenticated user after the AI stream completes.
Use the normal Monthly Tokens display to monitor account usage.
If Translation processing fails, the current interface displays:
Error
followed by the returned message or:
“Failed to translate. Please try again.”
The shared AI stream can also return:
“ReadableStream not yet supported in this browser.” when the required browser streaming functionality is unavailable.
Use a current supported browser.
Swap requires both:
If From remains Auto-detect, Swap stays unavailable.
Check:
Check:
Try:
AI Translation is generative.
You can:
Brand and product terminology can require manual review.
Correct protected or standardized terms in Edit before external use.
For:
use appropriately qualified human review before relying on the Translation.
Translation Hub is focused on translated content.
It does not establish exact preservation of the source document's:
especially when translating from images.
For long content:
For recurring business Translation, maintain an approved list of terms such as:
Then manually ensure those terms remain consistent across Translation results.
Before translating:
After translating:
The source verifies:
The current subscriber source does not establish:
TRANSLATION HUB
↓
Workspace:
AI TRANSLATION TOOL
↓
LANGUAGE SELECTION
↓
Choose:
FROM (OPTIONAL)
or:
AUTO-DETECT
↓
Choose:
TO *
↓
Enter:
ORIGINAL TEXT
or:
ADD IMAGE
↓
Optional:
QUICK EXAMPLE
↓
TRANSLATE
↓
TRANSLATING...
↓
Output:
TRANSLATION
↓
Choose:
PREVIEW
or:
EDIT
↓
Review and correct
↓
COPY TRANSLATION
Website Wiki transforms information extracted from a supplied website URL into a structured Wiki-style document.
The workspace description is:
“Transform any website into a structured, editable Wiki document.”
Navigate to:
Sidebar → Core Tools → Website Wiki
The route is:
/website-wiki.
The current Website Wiki route verifies that the user is authenticated.
If not, it redirects toward:
/login.
Like Translation Hub, the current Website Wiki page wrapper does not contain a visible FeatureGuard in the supplied source.
Actual Tool visibility can still depend on system configuration and Sidebar access.
Do not confuse:
Takes existing website content and converts it into documentation.
Creates and edits a new website.
Website Wiki does not provide Website Builder's visual web-design or publishing workflow.
The current client accepts a single URL and submits that URL to the extraction service.
It does not expose subscriber controls for:
The extraction backend is outside the sanitized bundle, so this manual does not claim that entering one URL automatically converts every page on the entire domain.
The verified workflow is:
Website URL
↓
Extract Raw Content
↓
AI Cleanup and Organization
↓
Editable Markdown Wiki
↓
Export
The primary field displays:
“Enter website URL (e.g., https://example.com/article)”.
If the URL field is empty, the exact current notification is:
Invalid URL
“Please enter a valid website address.”
Unlike the Live Chat URL importer from Chapter 21, the Website Wiki client does not explicitly check that the URL begins with http:// or https://.
It checks only that the URL field is not blank before submitting it to the extraction service.
For predictable results, enter a complete URL such as:
https://example.com/article
Pressing:
Enter
inside the URL field invokes the Generate Wiki workflow.
You can also select:
Generate Wiki.
Website Wiki performs:
Extract raw content from the supplied URL.
Send that extracted content to the selected AI Model for Wiki-style cleanup and organization.
The current client sends the URL to:
/api/website-wiki/extract
and expects:
rawMarkdown
in response.
For subscriber operation, this means Website Wiki first attempts to convert the source webpage into raw content suitable for further AI processing.
The actual extraction API implementation was excluded from the sanitized User Operations source.
Accordingly, this manual does not make unsupported claims about:
For predictable results, use content that can be accessed normally through a public webpage.
Website Wiki does not expose controls for entering:
After extraction, the raw Markdown is sent to the shared AI streaming endpoint using:
tool: website-wiki.
The Website Wiki system instruction tells the AI to transform raw website Markdown into a:
clean, well-structured, professional Wiki-style document.
The AI is instructed to keep the website's main:
The current instruction specifically tells the AI to remove:
The AI is instructed to organize the Wiki using:
headings.
It is also instructed to use:
where appropriate.
One of the explicit instructions is:
“Ensure technical details are accurately preserved.”
Even with that instruction, generated output should still be reviewed when technical accuracy matters.
If the raw website material is poorly organized, the AI is allowed to reorganize it logically.
Therefore:
Website Wiki is not simply a verbatim webpage copy.
It is an AI-refined documentation transformation.
The resulting Wiki may:
while attempting to retain the core informational content.
If you need a legally exact or archival copy of a webpage, Website Wiki is not the appropriate substitute.
It intentionally cleans and reorganizes the source.
The refined Wiki is generated through a streaming response.
As the AI produces text, the Markdown state is progressively populated.
After the processing sequence completes, Website Wiki displays:
Wiki Generated!
with:
“The website has been successfully converted into a Wiki using [selected Model]”.
If extraction or AI refinement fails, the current message is:
Process Error
followed by the returned Error or:
“Failed to generate Wiki.”
If the extraction stage fails, the current client can surface:
Scraping failed
when no more specific server Error is returned.
If the AI stage fails, the current fallback is:
AI Refinement failed.
If the browser does not provide the expected AI response stream, the current workflow can report:
Streaming not supported.
Before Wiki content exists, the empty state displays:
Ready to explore?
The interface says:
“Enter any website URL above to start extracting knowledge. Our AI will clean, structure, and convert the page into a beautiful Wiki format.”
Notice that this current UI wording specifically says:
the page
which reinforces the conservative user-manual interpretation of the Tool as URL/page-oriented rather than guaranteed full-site crawling.
The URL toolbar contains:
Test URL.
Test URL places:
https://en.wikipedia.org/wiki/Artificial_intelligence
into the URL field.
Selecting Test URL only fills the URL field.
It does not immediately launch the extraction/AI process.
Select:
Generate Wiki
afterward.
The empty state also provides:
Try with Wikipedia.
It performs the same basic action: setting the Wikipedia Artificial Intelligence URL as the current URL.
Once Wiki content exists, Website Wiki displays:
Select:
Copy
to copy the current Wiki Markdown.
The current notification is:
Copied!
“Markdown content copied to clipboard.”
Copy operates on the current markdown value.
Therefore, if you edit the Wiki first, Copy uses your edited version.
The actions toolbar contains:
Clear.
Clear removes:
The current Clear action does not show a source-verified confirmation prompt.
Therefore, preserve important Wiki content before using it.
No current:
workflow is established.
The generated document defaults to:
Preview.
Preview renders the Markdown as formatted documentation through the PROS SUITE Markdown Renderer.
Select:
Editor
to directly edit the underlying Wiki Markdown.
The current Editor placeholder is:
Edit your wiki content here...
Unlike some generated-output workspaces, Website Wiki gives you direct control over the generated document.
You can change:
before copying or exporting.
Exports use the current in-memory Markdown.
Therefore, manual changes made in Editor are included when you immediately export the document.
Direct editing does not mean the document is being persistently saved to your account.
The source does not establish an automatic Wiki project-save workflow.
The actions toolbar displays:
EXPORT AS
followed by three formats:
Select:
MD
to download:
website-wiki.md.
Markdown is the closest export to the actual Editor source.
It preserves the current plain Markdown content rather than converting it into a visual document format.
Select:
to generate:
website-wiki.pdf.
The current PDF exporter recognizes Markdown headings.
Its approximate Font sizing is:
using bold Helvetica.
Markdown lines beginning with:
-*are converted into bullet-style lines using:
•.
For ordinary PDF text lines, the current exporter strips **bold** Markdown markers and renders the contained text in the normal body style.
Therefore:
Inline bold formatting is not source-verified to appear as bold in the current PDF export.
The PDF export is designed for readable Wiki content.
It is not a pixel-perfect copy of the browser's styled Preview.
Review the exported PDF before external distribution.
The current exporter automatically creates additional PDF pages when content reaches the bottom margin.
Select:
DOCX
to generate:
website-wiki.docx.
The DOCX exporter recognizes Markdown headings and creates larger bold Paragraphs for them.
Unlike the current PDF path, the DOCX exporter explicitly recognizes:
**bold text**
and converts those sections into bold Word text runs.
The current DOCX exporter specifically handles:
It does not establish complete conversion of every Markdown feature into equivalent Word formatting.
Review complex documents after export.
The current default names are:
website-wiki.mdwebsite-wiki.pdfwebsite-wiki.docx
The current export controls call the exporter without providing a custom Filename.
Therefore, exports use the fixed default Names above.
Rename the downloaded file externally when needed.
For professional documentation, consider renaming the export after Download.
For example:
AIUNIFY_CRM_User_Wiki_2026.docx
or:
Product_Support_Wiki_August_2026.pdf
The Website Wiki page reads:
selectedModelisManualModelfrom the authenticated PROS SUITE context and submits them with the AI refinement request.
The Tool does not display a Model selector inside its own workspace.
Use the normal shared Header Model control before generating when Model choice matters.
Unlike Translation Hub's shared useGeminiStream helper, Website Wiki creates its own direct /api/ai/stream request.
That request includes:
but the current Website Wiki client does not add subscriber-configured provider API-Key headers.
Accordingly:
The current source does not verify that user-configured provider API Keys from Chapter 11 are forwarded by Website Wiki.
The selected Model is sent, but the client-side BYOK headers used by several other workspaces are absent here.
The generic feature catalogue currently lists:
Website Wiki — 20 Tokens.
Website Wiki does identify its AI request as:
tool: website-wiki.
However, the actual server-side Token-accounting route is outside the sanitized bundle.
Therefore, 20 Tokens is a source-verified catalogue/default reference but should not be described as an immutable live charge without current server/account confirmation.
The client first performs webpage extraction and then calls the AI refinement service.
The source does not expose separate user-facing Token prices for:
Do not invent separate charges.
Website Wiki does not display:
20 tokens per generation
inside the URL-generation interface.
Monitor Monthly Tokens through the normal PROS SUITE account controls.
Website Wiki currently holds:
in the current page session.
There is no source-verified database/project workflow for:
Because persistent Wiki Projects are not established:
Copy or export important content before leaving or refreshing the page.
There is no current:
field.
The generated document can contain Markdown headings, but the current UI does not establish a separate:
It is one generated editable Markdown document.
Website Wiki does not currently provide:
The supported outputs are copied Markdown and downloaded files.
These two systems should not be confused.
Website URL → AI-cleaned documentation file
Documents/URLs → Indexed RAG Knowledge Base → Customer-support answers
Website Wiki does not automatically train the Live Chat support Agent.
The current source does not establish a button such as:
Add Wiki to Live Chat Knowledge Base
If you want to use a generated Wiki as Live Chat training content, an appropriate manual workflow would be required, such as exporting a supported document and then adding it through Live Chat where appropriate.
Creates a new website.
Extracts information from an existing webpage and reorganizes it as documentation.
Researches and navigates webpages to answer an objective.
Transforms extracted webpage content into a structured document.
Browser Control produces research synthesis; Website Wiki produces editable Wiki-style Markdown.
Rather than entering a broad homepage when you need detailed documentation, use the most relevant page.
For example:
https://example.com/help/product-setup
may produce a more focused Wiki than:
https://example.com
Because full-site crawl depth is not verified, consider generating separate Wiki documents from important pages such as:
Then combine them externally when appropriate.
Check:
Use Editor to correct issues before exporting.
AI refinement can reorganize and rewrite extracted material.
Verify:
against the original webpage before publishing the Wiki.
The system is instructed to remove boilerplate, but extraction can vary by site.
Use Editor to remove anything that still looks like:
The most reliable workflow is:
Generate
→ Preview
→ Editor
→ Correct
→ Preview Again
→ Export
Because Markdown most directly preserves the editable source, exporting MD can be useful as the master text version.
PDF and DOCX can then serve as distribution/document formats.
The PDF converter has simpler formatting support than the browser Preview.
After export, inspect:
After DOCX export, inspect:
because the current DOCX conversion is intentionally basic rather than a full Markdown-to-Word layout engine.
Make sure the field is not blank and use a complete address such as:
https://example.com/article
Possible user-level causes include:
Try another accessible source page.
The current user interface submits one URL and provides no crawl-depth control.
Generate separate Wikis from the missing pages if necessary.
Compare it with the source webpage.
Then:
The system is explicitly instructed to remove boilerplate and reorganize content.
If useful information was mistakenly omitted:
This is expected.
Test URL only fills the Wikipedia URL field.
Select:
Generate Wiki
afterward.
Clear intentionally removes the current:
There is no verified Undo.
Export or Copy important work first.
Website Wiki does not currently establish persistent Saved Wiki Projects.
Use MD, PDF, DOCX, or Copy before leaving the page.
The current interface does not offer a filename field.
Rename the file after downloading it.
The current PDF exporter strips ordinary **bold** Markdown markers instead of rendering those inline sections in bold.
Use DOCX when preserving basic inline bold is important, or manually finalize the PDF externally.
The DOCX exporter applies its own basic Word formatting.
It is not a pixel-perfect export of the browser Preview.
Review and finalize the Word file before external use.
Before generating:
https:// address.After generation:
Before leaving:
The source verifies:
website-wiki Tool identification
The subscriber-facing source does not establish:
| CapabilityTranslation HubWebsite Wiki | ||
| Primary purpose | Language Translation | Website-to-document transformation |
| Text input | Yes | Extracted from URL |
| Image input | Yes | No |
| URL input | No | Yes |
| Source language selector | Yes | No |
| Target language selector | Yes | No |
| AI Model | Shared PROS SUITE Model | Shared PROS SUITE Model |
| Manual editing | Yes | Yes |
| Copy | Yes | Yes |
| MD export | No | Yes |
| PDF export | No | Yes |
| DOCX export | No | Yes |
| Saved Projects | No verified workflow | No verified workflow |
| History | No | No |
| Publishing | No | No |
TRANSLATION HUB
→ Select Source Language or Auto-detect
→ Select Target Language
→ Enter Text or Add Image
→ Translate
→ Review Preview
→ Edit
→ Copy Translation
WEBSITE WIKI
→ Enter Website URL
→ Generate Wiki
→ Extract webpage content
→ AI organizes raw Markdown
→ Review Preview
→ Edit Markdown
→ Copy or Export
→ Choose:
MD / PDF / DOCX
AIUNIFY PROS SUITE Translation Hub and Website Wiki are two specialized Core Tools for transforming existing information.
Translation Hub is presented as:
AI Translation Tool
and currently exposes 15 selectable languages, including English, Spanish, French, German, Italian, Portuguese, Russian, Japanese, Korean, Chinese, Arabic, Hindi, Dutch, Swedish, and Norwegian. The source language is optional and can remain on Auto-detect, while the target language is required.
Subscribers can translate either:
with a current image maximum of 5 MB.
Generated Translation is streamed into an output workspace containing:
The translated text can be manually corrected and then preserved with:
Copy Translation.
The current Translation Hub does not establish Saved Projects, Translation history, or file Download.
Translation Hub uses the shared PROS SUITE AI streaming system and therefore the shared selected Model and applicable provider API Keys. Although the generic feature catalogue lists Translation Hub at 10 Tokens, the specialized Translation page does not pass a Translator-specific Tool ID into the shared stream, so the final exact live Token charge cannot be proven from the sanitized user-side source.
Website Wiki is a separate workflow presented as:
“Transform any website into a structured, editable Wiki document.”
The subscriber enters a webpage URL and selects:
Generate Wiki.
The current client first requests raw website content and then sends that extracted Markdown into an AI refinement stage. The AI is instructed to:
The resulting Wiki provides:
exports.
The current default export filenames are:
website-wiki.mdwebsite-wiki.pdfwebsite-wiki.docx
Website Wiki uses the shared selected AI Model, but its specialized client request does not currently forward the subscriber-configured provider API-Key headers used by several other PROS SUITE AI workflows. The generic catalogue identifies Website Wiki with a 20-Token reference value, but final server-side billing remains outside this User Operations source.
Neither Translation Hub nor Website Wiki currently establishes persistent Saved Projects. Therefore, the correct operating principle for both systems is:
Generate → Review → Edit → Preserve the Final Output Before Leaving
For Translation Hub:
Copy the reviewed Translation.
For Website Wiki:
Copy or export MD, PDF, or DOCX.