MetaServer > Version History

Here you will find all MetaServer release notes, including details of new features, improvements and fixes.

You can always download the latest version of MetaServer including Operator, Admin clients and sample workflows on the MetaServer Product Page. If you are looking for base installers without any sample workflows, please refer to our Downloads page.

If you wish to get notified as soon as a new version of MetaServer is published, please subscribe to our CaptureBites Newsletter.

IMPORTANT: Before refreshing or updating your MetaServer, please pause your MetaServer first. You can do this in your Admin Client, under the Server tab. As soon as all your action queues are "yellow" (= paused), you can perform your refresh or update.

This will ensure that no documents in your current queue become corrupted during your refresh or update.

Version 3.1(33) | 2025-11-28

NEW #2238: EXTRACT TEXT – TEST PREVIEW: We now show the extracted result of an extracted word group as a tooltip when you hover over an extracted pink rectangle in the viewer.

The tooltip shows the following information:

• Extracted value
• Data type: Printed, Handwritten, Barcode
• Confidence level in %
• Font size in pt

When the tooltip pops up, you can right-click and copy the value in the clipboard by selecting “Copy Tooltip to Clipboard”.

NEW #2230: CONVERT TO SEARCHABLE PDF – AZURE AI DOCUMENT INTELLIGENCE: You can now also select an “Azure AI Document Intelligence” resource in the Convert to Searchable PDF action. Previously there were only two choices: “Tesseract” and “Azure AI Vision”.

NOTE: The “Azure AI Vision” engine will be retired on 25 September 2028. The “Azure AI Document Intelligence” engine will be replacing it.

NEW #2251: IMPORT EMAIL – REPAIR PDF FILES: This new option in the Import Email action, converts problematic PDF attachments into single-layer image PDFs for full MetaServer compatibility and creates a “CBMSWarning” log file. It saves the corresponding file in:
C:\CaptureBites\MsData\Log

The option is disabled by default.

ENHANCEMENT #2092: OFFICE 365 EMAIL: Increased security by using app registration instead of a secret-based connection. MetaServer will be registered as “CaptureBites MetaServer” in Microsoft Entra ID Enterprise Applications.

The Office 365 Email connections must be renewed and the installer will show the following warning:

“The Office 365 connections must be renewed. Do this by launching the Admin client.” 

When starting the Admin client after installation of this version, you will get a list of email accounts that need to be renewed.

ENHANCEMENT #2226: VALIDATE – DATABASE: Previously, for the Validate Database rule, we only sorted on the first column of the lookup list.  We now sort the lookup list on column 1 and column 2, to make it easier to select the correct record.

ENHANCEMENT #2247: SERVER – CORES: CONVERT TO SEARCHABLE PDF: We have added the “Convert [ X ] pages in parallel’“ option (speeds up the conversion of PDFs with many pages) to the Cores manager in the Server tab and removed it from the Convert to Searchable PDF action. This centralizes the allocation of cores to all actions in a single screen.

ENHANCEMENT #2261: SERVER: DOCUMENT DATA STRUCTURE: We now only keep the original document set once for each separated document. With large sets separated in many documents, the disk usage was very high previously. This is now considerably reduced.

ENHANCEMENT #2262: ADMIN CLIENT: SETUP SCREENS: We have disabled the scroll wheel on selection lists and spin boxes. Especially in the Mark Detection rule’s setup, where not all controls are visible, people sometimes scrolled down and when their cursor was over a selection list or spin box, they changed the value of the selection field or spin box accidentally.

ENHANCEMENT #2263: ADMIN CLIENT – WORKFLOWS: FIELDS: In the Workflow tab’s Fields setup window, we have added an “Add” button to insert a field anywhere in the list of fields. Previously, you could only add a field at the end of the list and then move it up.

FIX #2225: WINDOWS DISPLAY SCALING WAS IGNORED: With very large and high-resolution screens becoming more popular, users often set the display scaling to 125% or 150%. This sometimes caused the Operator and Admin client to stay in the original scale and consequently appeared very small with hard to read fonts.  We now fully support Windows display scaling, and the Operator and Admin client will increase size accordingly.

FIX #2233: SERVER – AZURE RESOURCE MANAGER: When you copied a workflow directly in the data folder using an Azure Resource that did not exist on the system, the following error occurred when trying to adjust settings within an action or rule that used this Azure AI resource:

“The given key was not present in the dictionary.”

You could not open the setup, so it was impossible to change the Azure AI resource to one that exists on the new system. Now, the setup opens and you can switch to an existing resource.

FIX #2255: SERVER – AZURE RESOURCE MANAGER: Copying Azure Resource Settings to another system failed and returned the following error:

“The given key was not present in the dictionary.”

FIX #2229: SEPARATE DOCUMENT / PROCESS PAGE(S): During the Separate Document / Process Page(s) action, if you used the Extract Text (Azure AI Vision) or Extract Text (Azure AI Document Intelligence) rule, an error showed during runtime:

“The Azure resource manager is not yet instantiated”.

FIX #2275: ORGANIZER – CLASSIFICATION MODE: The “Always check” option always added the first document to the training set, even if it was valid (= green) and the document type was not changed.

FIX #1848: EXTRACT – MARK DETECTION: The Mark Detection rule caused errors when switching on the Deskew, Rotate and/or Convert to Seachable PDF option in any Extract Text rule (TOCR, Azure AI Vision or Azure AI Document Intelligence) while setting up scaling and registration.

FIX #1130: EXTRACT – MARK DETECTION: If none of the anchor words were detected in the Mark Detection rule, the Auto-rotate and Deskew options, as set in your Extract Text rule, were not applied.

FIX #1107: EXTRACT – MARK DETECTION: If none of the registration points were found, the rule did not return the default value. Now, it returns the default value at 0% confidence instead of nothing.

FIX #2257: EXTRACT – MARK DETECTION: The Admin client showed an Out of Memory error after pressing TEST Scaling Points or TEST Registration Points many times.

FIX #2258: EXTRACT – MARK DETECTION: There was a display issue in the test viewer when a scaling point fell outside of the image frame.

FIX #2255: EXTRACT – MARK DETECTION: There was a display issue in the test viewer when a scaling point fell outside of the image frame.

FIX #2236: EXTRACT – ELECTRONIC PDF: On very rare occasions (we only encountered this case with 2 documents in the last 10 years), when extracting text from the electronic PDF text layer, some text was returned double. We now detect this anomaly and correct the text automatically.

FIX #2243: EXTRACT – FORMAT CSV: Merging columns with “highest row count” deleted columns when they did not match the highest row count. This caused problems with the number of columns in the final CSV.  Columns that do not match the highest row count are now preserved with blank values instead.

FIX #2224: EXTRACT – REPLACE TEXT: Replacing the “₽” character (= Rouble) or ₶ – (= Livre Tournois) replaced every character in the field.

11/2024
02/2025
04/2025
06/2025
09/2025

But it did accept:

10/2024
12/2024
01/2025
03/2025
05/2025
07/2025
08/2025

FIX #2234: VALIDATE CSV: The validation client showed an error when you deleted more than one value in the table.

FIX #2247: VALIDATE CSV – DB LOOKUP IN A COLUMN WITH FILTER:

1) The filter did not work when using a column value.

2) Filters did not work with regular field values, only other columns could be selected.

FIX #2260: VALIDATE – OUT OF MEMORY ERROR: Validating many documents where a lot of “Select Text (OCR Tool)” activity was applied, eventually showed an “Out of Memory” error.

FIX #2235: CONVERT TO MULTIPAGE TIF: An error occurred when processing a large document of 4693 pages:

“An attempt was made to move the file pointer before the beginning of the file.”

This happened when the file reached the max file size of our TIF writer, a signed 32-bit value of 2 GB. We changed it to an unsigned 32-bit value of 4 GB. This should make it possible to process TIFs around 9000 pages.

If a TIF file reaches 4 GB, the conversion fails, and we report a meaningful error:

“TIF file size exceeds 4 GB.”

FIX #2205: CONVERT TO PDF MRC: The Convert to PDF MRC action can compress pages with both sides of the images being less than 28 inch (71,12 cm). When a page size exceeds this limit (Error 56), we skip conversion of that page but still convert all other pages that do not exceed the 28 inch limit.

FIX #2277: EXPORT TO SHAREPOINT: We have enabled chunking to be able to upload very large files (> 30MB).

FIX #2240: TASKS – SYNC METASERVER DATABASE: In some cases, the MetaServer database was not re-opened after sync and caused an error when a rule tried to access the database:

“Object reference not set to an instance of an object.”

We now force a reopen of the database and log to a “CBMSWarning” file:

“Database has been reopened: {DatabaseFileName}”.

You can find this log in:

[ Data Drive ]\CaptureBites\MsData\Log\[ Year ]\[ Month ]

FIX #2256: TASKS: The trigger “When MetaServer is running for [ X ] minutes” ran the task every 15 seconds instead of only once.

FIX #2239: VALIDATE CSV – HELP: In the Validate CSV setup’s Database column setup window, the “Help” button now refers to the correct URL:
https://www.capturebites.com/metaserver/help/validate/040-020/

Version 3.1(32) | 2025-09-30

NEW #2018: SEPARATE & CLASSIFY: The Separate & Classify action makes use of an Azure AI Document Intelligence Custom Classification model to separate a document set in multiple documents and also to allocate a document type and confidence level to each separated document. This new action goes hand in hand with the new “Classify” mode in the Organize action (see #1999).

NEW #2021: PRE-ORGANIZE: Since we have implemented the Azure AI Classification feature (see previous note), we have introduced a “Pre-organize” action to generate classification separators. They are marked in different colors:

Very confident (green)

Less confident (orange)

Low confidence (red)

The “Pre-organize” action also pre-renders the document thumbnails which accelerates opening a document in the Organizer considerably.

NEW #1999: ORGANIZE: CLASSIFY MODE: With the Organizer’s new Classify mode (as opposed to the default Organize mode), you can show each set’s document type separators to easily check if each document was correctly classified.

The Classify mode is typically combined with the new Separate & Classify action (see #2018).

The document type separators show the document type and the confidence level.

The separators are colored differently depending on the confidence level of the detected document type. The confidence thresholds can be set in the Organize action:

ORANGE: Separators below a set confidence level (e.g. 50%) are orange.

RED: Separators below a set confidence level (e.g. 15%) are red.

GREEN: Separators above the orange threshold are green. Also, if the operator manually confirms or corrects separators, it becomes green. Only when all separators are green, the user can OK the documents.

The available document types are listed in a drop-down list in an indexing panel. You can use it to change the document type or confirm the detected one.

The document types can be directly retrieved from the classification model or you can store them in a MetaServer database (CSV).

IMPORTANT: If you use a MetaServer database to store your document types, the first column of that database is automatically considered as the column containing the document types.

NEW #2124: ORGANIZE: CLASSIFY MODE: In the Organizer’s Classify Mode, if a document is misclassified (= red / orange), or if it is correctly classified but questionable (= orange), and you correct or confirm the document type, that document is added to the “Corrected documents” list. This list also has a check mark to signify the corrected documents to be added to an incremental training set and update the custom classification model.

The incremental document set is stored in:
C:\CaptureBites\MsData\Training\Classification\[WORKFLOW_NAME]

The file is named:
[CORRECTED DOC TYPE]_[OPERATOR NAME]_[DATE AND TIME]_(detected_[DOC TYPE BEFORE CORRECTION]_confidence_[CONFIDENCE PERCENTAGE]).PDF

For example:
FACTURE_johndoe_2025-09-08_12.30.00_(detected_ATTESTATION_confidence_15%).PDF

NEW #1982: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE) – CUSTOM EXTRACTION: With the new Custom Extraction option for the Extract Text (Azure AI Document Intelligence) modules, you can build your own custom extraction models.

If you have documents for which there is no prebuilt model, like shipping documents, litigation documents, HR documents, contracts, etc. you can now create your own custom extraction model.

The concept is straightforward:

1) You collect at least 5 representative samples of the document. These samples will be used to train the Custom Extraction model. You also need several other samples, different from the training samples to test the model.

2) You build your custom extraction model using the Azure Document Intelligence Studio (DI Studio).

In the DI Studio, you create a new “Custom Extraction” project. For example, “Car Lease Agreements”. Next, you define the custom fields you want to extract, like “Lessor”, “Lessee”, “Effective Date”, “End Date”, “Leasing Price”, “Car Make”, “Car Model”, “VIN”, etc.

3) Next, you upload a minimum of 5 different training samples of your documents and, just by selecting and clicking, you link each field with the correct data on the samples.

4) Once you have finished this process, you train the custom model.  After training is complete, your custom model becomes available in MetaServer, and you can select it from the “Model” drop-down list of the Extract Text (Azure AI Document Intelligence) rule.

5) Finally, you map MetaServer fields with the custom fields of your custom model and it’s done. You can then test the model with fresh samples that weren’t used in your training set. If necessary, you can add additional samples to your training set to improve recognition quality. A model trained with most, if not all, variations of your custom document will result in a highly reliable and impressive extraction model and produce excellent data output.

To make use of Custom Extraction, you need to license the following 2 MetaServer modules:

1) CB-META-AZDI: Extract Text (Azure AI Document Intelligence)

2) CB-META-AZCE: Extract Text (Azure Custom Extraction)

You’ll also need to sign up for an “Azure AI Document Intelligence” resource through the Azure Portal, if you haven’t already done so.

Microsoft’s pricing for the Azure AI Document Intelligence engine and options can be found here:
https://azure.microsoft.com/en-us/pricing/details/ai-document-intelligence

NEW #2019: TASKS – SYNC METASERVER DB – CUSTOM CLASSIFICATION MODEL: You can now synchronize a MetaServer database with the document types of an Azure AI Document Intelligence Classification model with the Sync MetaServer Database Task. It is typically used in combination with the new Organizer’s “Classify” Mode (see #1999).

NEW #2186: EXPORT TO SHAREPOINT – OAUTH20: On August 19, 2025, Microsoft stopped support for basic authentication. Our Export to SharePoint module used basic authentication and stopped working.  This version fixes the problem and changes to Oauth20 authentication, also known as “modern authentication”.

The connector needs to be re-logged in using the new method.

The first login will always fail because the Azure Entra ID admin needs to give permission to give the connector access to SharePoint, triggering the following warning:

If you cannot log in like this, you will need to go to the Entra Admin Center en set the permissions there.

You can find a detailed guide on how to grant “Application Access Rights” to the Export to SharePoint module here:
https://www.capturebites.com/metaserver/help/0auth20-sharepoint/#01

NEW #2211: IMPORT EMAIL – WHITELIST: In the Import Email action, we now support whitelisting sender emails. Emails from non-whitelisted email addresses will not be imported and will remain unread in the watched inbox.

The whitelist is maintained in a MetaServer database (= CSV).  Just select a MetaServer database where the first column contains the whitelisted senders.  The column header can be anything we just consider the first column as the whitelist.

To disable this option, don’t select any whitelist.

In your database, you can also make use of wildcards like “*” and “?”.

Asterisk (*): Matches zero or more characters.

as in: *moons@capturebites.com

=> accepts all capturebites.com emails with the last name “moons”

like: jdmoons@capturebites.com or sophiemoons@capturebites.com

It will not accept an email address like:
phboons@capturebites.com

Question Mark (?): Matches a single character.

as in: ap_??_region@capturebites.com

=> accepts all capturebites.com emails starting with “ap_”, followed by any 2 characters and ending with “_region”

like: ap_fr_region@capturebites.com or ap_de_region@capturebites.com

It will not accept an email address like:
ap_emea_region@capturebites.com
ap_n_region@capturebites.com

If you regularly need to update your MetaServer database containing whitelisted emails or import it from another location, you can set up a Task to automate this.

You can find more detailed information on the following online help pages:
Import a MetaServer DB
Sync a MetaServer DB

NEW #2133: IMPORT FROM FOLDER – IMPORT LIMIT: The “import limit” in the Import from Folder action applies to documents placed in the root of the watched folder and each subfolder of the watched folder individually.

For example, you are importing from a folder that contains 10 subfolders. If you set an import limit of 25, you can have 25 documents in the root folder and 25 documents in each subfolder (= total of 250 documents). All 275 documents will be imported.

If a subfolder receives a 26th document, a slot needs to be freed up first from another subfolder to get it imported.

In comparison, the “import limit” in the Server tab is an absolute limit, meaning that all importing will be paused when that limit is reached. It will resume importing as soon as a slot becomes available again.

The use-case for this option is to divide the imported documents over several departments, each one scanning into a different subfolder. By setting the limit at subfolder level, we make sure that a single department does not take the full import limit hostage.

NEW#2079: SERVER: NUMBER OF CORES: We have switched to a centralized core management function in the Admin Client’s Server tab. This replaces the “Number of cores” used in in the setup of the following actions:

– Convert To PDF MRC
– Convert to Searchable PDF
– Kofax VRS
– Export to Box
– Extract
– Separate Document / Process Page

The new centralized cores setup features a drop-down menu to reset the cores configuration to the default setting or to set all core settings to 1 to process documents in the sequence as they were imported.

Your originally configured number of cores from each individual action will automatically be converted to the centralized Server Core settings. 

You can now also adjust the number of cores for the following actions:

– Convert to Multipage TIF (default: 4)
– Preprocess Document (default: 4)
– Scale Pages (default: 4)

IMPORTANT: The number of conversion cores determines how many documents will be processed in parallel for each action. If you add up all the cores of all actions, you get the total number of cores. We don’t advice to set the total number of cores higher than the number of physical cores available in your computer since it can lead to a system overload.

If you want to calculate the number of physical cores required for your processing needs, please refer to our online physical core calculator.

NEW #2132: SERVER: CENTRALIZED AZURE RESOURCE MANAGEMENT: You now only need to define an Azure Resource once by using the Azure Resources setup in the Server tab.

You can then use that resource from any Azure-related action or rule by selecting it from a drop-down list. If the resource’s key changes, you only need to change it once and all rules using that resource will use the new key.

There are 3 types of Azure resources you can create:

1) Azure AI Vision (reads printed and handwritten text)
2) Azure AI Document Intelligence (reads printed and handwritten text, barcodes and uses pre-built and custom models)
3) Azure AI Language (summarizes text)

This change affects the following:

Actions:

– Convert to Searchable PDF
– Separate & Classify
– Organize

Rules:

– Extract Text (Azure AI Vision)
– Extract Text (Azure AI Document Intelligence)
– Format (Azure AI Summarize)

Tasks:
– Sync MetaServer Database

When you update to this new version of MetaServer, your existing workflows will be automatically converted to use the centralized Azure Resource setup.

– Any Azure resources used in your workflows and tasks will be automatically created.

– When you still have documents in the queue, embedded workflows will use the central Azure resource(s).

– Azure Resources are saved as *.CBMSResource files in the following location:
C:\CaptureBites\MsData\Resources

The resource’s key is encrypted.

– Before an Azure Resource gets deleted in the centralized setup, we will first check if it is still in use by 1 or more workflows or tasks. Only Azure Resources that are not linked to an existing workflows or task can be deleted.

– When you select an Azure Resource, you can see a list of all the associated workflows and tasks.

NEW #2132: VALIDATE – REJECT: We have implemented an option where, when a Validate action’s reject purpose is positive (= green), you can force the operator to check the validity of the fields.

This is useful when you need to use the Reject function as a positive action like, for example, setting aside documents for updating a Custom Classification model.

In that case, documents are processed like regular documents and set aside in a special “training set” folder. The fields of the document should still be correct.

To apply this method, you just need to select a green icon for the Reject function (e.g. the “Save to Folder” icon, “Loading” icon, “Green alien” icon, etc.).

NEW #2068: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE) – CHECK (US) MODEL: For the Extract Text (Azure AI Document Intelligence) rule, we now support the prebuilt “Check (US)” model. It can also be used with non-US checks but, with non-US checks, the MICR line is not detected and needs to be found within the full text using conventional MetaServer rules.

ENHANCEMENT#2081: DOCUMENT COUNTERS: We have added additional info about the number of sets and documents in workflows, CPU, memory, disk usage and disk capacity.

The usage % appear in orange or red when reaching the following levels:

ENHANCEMENT #2082: LOG – CPU USAGE: The CPU usage is now monitored. As soon as your system is 100% busy for more than 5 minutes, the monitor waits until the CPU usage drops below 100%. It then writes a warning log in the log folder located in:
C:\CaptureBites\MsData\Log

For example:

If the CPU is 100% in use for 5 minutes since 13:00, and then only drops down below 100% after another 10 minutes, the warning log with detailed information is written to disk on 13:15.

In case of a CPU overload, your computer becomes slow and unstable. We recommend monitoring the CPU usage in the “Performance” tab in Windows’ Task Manager when MetaServer is processing documents.

If the CPU is regularly hitting 100%, we recommend decreasing the number of cores. This is configured in the Server tab of the MetaServer Admin client by pressing the “Cores” icon. Decrease the number of cores until the CPU usage stays below 100%.

As a reference, the most processor-hungry actions are:

– Extract
– Separate Document / Process Page
– Kofax VRS
– Convert to Searchable PDF

ENHANCEMENT #2180: EXTRACT BARCODE: UPGRADE TO LATEST SDK 9.4.1: The main benefit of this new Extract Barcode engine version is an increase in speed of up to 20% compared to the previous 9.3.1 version using identical settings.

There is also significant improvement to the recognition of barcodes at multiple angles on the same page.

ENHANCEMENT #2212: VALIDATE: USER-LEVEL SETTINGS: Previously, settings like font size, filters, layout etc. were saved at device level and were the same for all users. This gave unexpected results when some users preferred other settings than their colleagues. We now save the client settings at user-level.

ENHANCEMENT #2134: ORGANIZE / VALIDATE: OPEN DOCUMENT LIST: The filter of the Document List was not applied during Organize or Validate. Now, you can apply a filter and the Organizer or Validation will only present the filtered documents.

If you run out of filtered documents, you can go back to the “Document List” to adjust the filter to continue organizing / validating the remaining documents.

The column header of a filtered column is also now marked in green.

ENHANCEMENT #2152: ORGANIZE / VALIDATE – OPEN DOCUMENT LIST: We have optimized list-loading for reserved documents and filtered lists.

We have changed the method to list the available documents for the logged in user. The server will now only return unreserved documents and documents reserved for the current user and won’t even load unavailable documents anymore. This speeds up document list creation and it also avoids the problem that the list-limit potentially returns nothing even if some documents are reserved for the logged in user.

We do a similar thing with the filter (see #2134). We will only retrieve documents matching the selected filter. This speeds up filtered document list creation considerably.

ENHANCEMENT #2107: RESERVE DOCUMENTS FOR A USER GROUP: If the selected user group contains sub-user groups, then the users of the sub-user groups can also access the reserved documents.

For example, if the user group “MetaServer Operators” contains 3 users, John, Joe and Jim and the” Administrators” user group reserves a document for the “MetaServer Operators” user group, it can also be accessed by any member of the “Administrators” user group.

WARNING: If you make changes to Active Directory user groups, you need to log out your windows session and log in again to apply the changes.

WARNING: The logged in user will only see groups where they are a member of. If you want to browse user groups from the Reserve rule, you need to make sure that, whoever is setting up the Reserve rule, is member of the same user group. Alternatively, they can enter the group’s name manually.

ENHANCEMENT #2030: MONITORING: We have added entries for reserved documents per user / group in the JSON monitor file.

This can be used in combination with monitoring software to send an email alert to a group with a certain number of documents in their validation queue.

For example, assume there would be 5 documents reserved for jdmoons, 10 for the “MetaServer Operators” active directory group and 12 in Organize for the “MetaServer Organizers” group. The JSON entries would look like this:

Organize (MetaServer Organizers): 12

Validate (jdmoons): 5

Validate (MetaServer Operators): 10

ENHANCEMENT #2135: IMPORT FROM WEB SERVICE: We added an Enable / Disable toggle to the Import from Web Service action’s setup. Disabling the import will delist the workflow from the web client’s Upload screen

ENHANCEMENT #2052: VALIDATE / ORGANIZE – SHOW GRAPH: The selected icon for the Validate or Organize action’s Reject function now sets the color in the workflow for the “rejected” flows.

For example, when you use a green folder icon, the graph will show the connected action as green.

ENHANCEMENT #2141 #2174: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE): We now expose custom models’ expiration date, field type and field description in the setup.

ENHANCEMENT #2147: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE – PREBUILT MODELS: We now retrieve all fields from the prebuilt models with a GET MODEL call, assuring that we give access to all the fields of the prebuilt models.

As a consequence, the “Invoice” model now also returns the tax details for all VAT rates similar to the receipt model:

– Tax Details Amount
– Tax Details Currency Code
– Tax Details Currency Symbol
– Tax Details Rate

ENHANCEMENT #2140: CONVERT TO PDF MRC: To convert large plans to PDF MRC, they were reduced in pixels to the largest rendering DPI. This made the image smaller than 17.5 megapixels and preserved dimensions. The DPI was limited to fixed steps of 100, 150, 240, 300,… DPI.  So, if 150 DPI was just above 17.5 megapixels, the image became 100 DPI. This method made it lose a lot of pixels.

Two changes have been made:

1) We scale to more pixels: Large plans are now reduced in pixels to the largest rendering DPI. This maskes the image smaller than 35 megapixels.

2) The DPI is set at the unit level: If, for example, 150 DPI does not match 35 megapixels, it will apply whatever DPI brings the image just below 35 megapixels (e.g. 143 DPI).

ENHANCEMENT #2182: OAUTH20 LOGIN FOR OFFICE 365 + BOX: When the WebView2 runtime was not installed and you tried to log in with Oauth20, you got a cryptic error.  We now show a meaningful error including a link to the WebView2 runtime installer.

ENHANCEMENT #1869: EXPORT TO SAP: We have changed from “KGS Activator Lite” to “Arcana X-Connect”. The connector does not require a transport anymore but registers with SAP through a certificate. This means it can update the content management system directly.

FIX #2183: IMPORT FROM FOLDER – CONVERT XFA TO IMAGE PDF: The Convert to XFA to Image PDF option correctly converted PDF XFA to Image PDF, but the check before processing the PDF through actions that don’t support XFA still failed and considered the PDF as XFA. This is now fixed.

The below actions, divided in groups 1, 2 and 3, were failing to process XFA files.

Group 1: Actions that failed if PDF is XFA/Text, process if XFA/Image

– Kofax VRS

– Set PDF Properties

– Convert to Black and White

– Convert to PDF MRC

– Convert to Searchable PDF

– Delete Page

– Add External Documents

– Split Booklets

– Digital Imprinter

Group 2: Actions that always processed documents:

– Convert to JPG

– Convert to Multipage TIF

Group 3: Action skipped if PDF is XFA/Text, processed if XFA/Image:

– Scale Page(s)

FIX #2087: IMPORT EMAIL: We have fixed the preprocessing error “Could not find a part of the path”. We now do an automatic retry in case this error appears.

FIX #2204: IMPORT ACTIONS – FIT TO PAGE SIZE OPTION: When the “Fit to page size” option was enabled, it changed the document’s pixel count. Now, it only changes the dimensions by setting the DPI correctly.

NOTE: The purpose of this option is to set the page size of smartphone images to A4 or letter size. These often feature the wrong 96 DPI or no DPI, turning them in gigantic dimensions.

We now force these images to become A4 or Letter size.

FIX #2213: IMPORT EMAIL: Embedded images in the email body with very long file names, exceeding the Windows file name length limit, caused the email to block the email import queue. We now handle this issue correctly.

The specific case was related to an email with a logo file named:
“OutlookEmoji-Ein Bild, das Schrift, Grafiken, Logo, rot enthält. KI-generierte Inhalte können fehlerhaft sein.b474f5c6-46c8-47a2-a8dd-a2f9e9530e8e.png”

FIX #2083: CONVERT TO PDF/A: Converting a 2564-page document caused an OOM. It now works after moving the action in an external .EXE.

FIX #2099: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE) – OTHER DOCUMENT MODEL: Line item fields did not return confidence levels.

FIX #2104: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE): The “Read” and “Other Form” model did not return the dominant language.

FIX #2118: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE): For the “ID Document” model, the Machine Readable Zone (MRZ) subfields now have a pink rectangle equal to the MRZ zone.

FIX #2137: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE): Vertical text objects were highlighted incorrectly with the wrong pink rectangle. Pink rectangles are used for highlighting the extracted data during validation.

FIX #2216: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE) – CUSTOM EXTRACTION: We only supported dynamic tables. Dynamic tables have a variable number of rows with tables potentially spilling over several pages. We now also support fixed tables.  Fixed tables have a fixed number of rows. If any rows are empty, they will also be returned.

FIX #2218: EXTRACT TEXT (AZURE AI DOCUMENT INTELLIGENCE): We now sort custom models alphabetically in the model selection list.

FIX #2184: EXTRACT TEXT (AZURE AI VISION): Erratic Error fixed with Azure AI Vision:

Value cannot be null. Parameter name: requestUriString

This is now treated as an internet outage. The relevant Extract or Convert to Searchable PDF action turns orange and an automatic retry is performed until the error does not occur anymore and the action will revert to green.

FIX #2080: EXTRACT – FORMAT CSV: The Table View of “;” delimited CSVs did not work. Only “,” delimited CSVs were correctly displayed in the table view.

FIX #2185: EXPORT PROCESSED SET: When a set of multiple separated documents was processed, the Export Processed Set action did not always email all separated documents.

This was related with the way we detect when a set is completely processed. If the Apply Separation & Page Processing action was still working on a set while some separated documents of the same set were already processed, the issue occurred.

We now stop all documents of a set for further processing until the Apply Separation & Page Processing action has completely finished, and all document of the set are available for processing.

FIX #2098: SERVER – DATA DRIVE: After moving all files to another drive, the following error appeared:

Failed to Start Server when moving the server data folder.

The server had to be stopped and restarted using the external shortcuts, and, although all files had been moved to the new drive, the MetaServer still used the previous drive. This has been fixed.

FIX #2184: EXPORT TO DATABASE: An erratic error has been fixed. “ExecuteNonQuery” requires an open and available connection. The connection’s current state is closed. This error is now automatically retried to avoid documents ending up in the Errors tab.

FIX #2197: EXPORT TO DOCUWARE: We now auto-retry the error: “The underlying connection was closed,…”

FIX #2189: HELP BUTTON URLS: The following actions’ Help button URLs have been corrected:

– Apply Separation & Page Processing

– Apply Page Processing

– Convert to JPG

Version 3.1(31) | 2025-03-27

NEW #1971: EXTRACT (AZURE AI DOCUMENT INTELLIGENCE) – CUSTOM CLASSIFICATION MODEL: We implemented Azure AI Document Intelligence’s Custom Classification model. You build your classification model using the Azure Document Intelligence Studio (DI Studio).

You can find a detailed guide on the Extract Text (Azure AI Document Intelligence) help page.

In a nutshell, it works like this:

1) In the DI Studio, you create all your document types and feed 5 or more representative samples for each document type.

2) You train the classification model. After it has completed building the model, it becomes available in the “Model” drop-down of the MetaServer Extract (Azure AI Document Intelligence) rule.

3) You select the classification model you want to use and you can then map the “Document Type” field with a MetaServer field.

The model also returns a confidence level, which you can use in Validation to check and, if necessary, correct low-confidence classifications.

To enable Custom Classification models in MetaServer, you require a new module with product code “CB-META-AZCC” (= Azure Custom Classification) on top of “CB-META-AZDI” (= Azure AI Document Intelligence).

NEW #2006: EXPORT TO COTRANET: With the Export to Cotranet action, you can export documents to Cotranet Cloud or On-Premise.

You can select and dynamically create the folders and subfolders where you want to store your documents. You then map MetaServer data with the selected Document Type’s index fields and name the documents and files independently using fixed values and/or MetaServer fields.

For more info regarding the Cotranet platform, please visit www.cotranet.com.

NEW #2014: VALIDATE – CREATE FAST DISPLAY DOCUMENT: We have added a new option called “Create fast display document”. A fast display document contains reduced resolution versions of very large pages (> 17.5 Megapixels).  The file is temporary and for display purposes only. It will not affect the output.

NEW #2029: SERVER MONITORING – MEMORY CONSUMPTION: In the Admin Client’s Server tab, we have added a new option called “Memory Consumption”.

When enabled, the file name of the memory consumption is pre-defined as:

YYYY-MM-{Server Computer Name}-Memory.CSV

This will create a new memory log every month.

It registers all important memory information, which will help with diagnosing systems that show memory issues and unexpectedly unload the MetaServer Service.

FIX #2012: ERRATIC RETRY ERRORS: The following errors were not retried correctly;

“Unable to delete file…”

“Impossible to create an existing file.”

On the client side, during setup, the information is saved in:
C:\ProgramData\CaptureBites\Programs\Admin\Data\Log

On the server side, the action queues save the information in:
C:\CaptureBites\MsData\Log

Affected actions / rules:

• Convert to PDF/A
• Convert to Searchable PDF
• Export to Alfresco
• Export to Box
• Export to Cotranet
• Export to DocuWare
• Export to Enadoc
• Export to SAP
• Export to SharePoint
• Export to WebServer:
  • APE
  • Gouw7
  • InProces
  • ZaakSysteem
• Extract Text (Azure AI Document Intelligence)
• Database Lookup
• WebHook

FIX #2026: EXPORT TO FOLDER: When the “Export files” option was disabled, you still received errors regarding mistakes in the definition of the files. We don’t show these irrelevant errors anymore, if the option is disabled.

FIX #2038: TASKS: When you create a task with a trigger at a fixed hour, it also executed the task instantly after publishing.

Now, when you publish a new task, it only runs when the trigger occurs.

FIX #2011: FORMAT CSV: Merging a CSV with repeating single line fields did not work correctly anymore. The internal code of the Format CSV rule is refactored and easier to maintain.

This enhancement was published in version 3.1.31.68 (2025-03-27).

FIX #2069: FORMAT CSV: In the Format CSV rule, we added a tooltip to the “Delete row if…” option and updated our online help page:

 “With this option you can choose if you’d like to delete rows from your CSV in certain conditions. Leave the condition blank if you don’t want to delete any rows.”

For example, you can delete rows when a complete row is empty. Or you can delete rows if a specific field, like “Total Amount”, is empty.

Special cases:

When a CSV is added to the columns in your setup:

– You can only delete rows when a complete row is empty. The other conditions, “if a specific field is blank”, won’t be applied.

– If the CSV is combined with repeating single line fields and the “Delete row if complete row is blank” is enabled, then first the blank rows are deleted from the CSV, next the repeating single line fields are added.

This enhancement was published in version 3.1.31.68 (2025-03-27).

FIX #2070: EXTRACT (AZURE AI DOCUMENT INTELLIGENCE): In the Extract Text (Azure AI Document Intelligence) action, some models crashed on blank PDFs. The “Invoice”, “ID Document”, and “Receipt” models failed on a totally bank PDF. The “Read” and “Other Form” models did not fail.

This enhancement was published in version 3.1.31.68 (2025-03-27).

FIX #2078: CONVERT TO MULTIPAGE TIF / CONVERT TO IMAGE PDF /CONVERT TO JPG: These convert actions occasionally reduced the resolution of some pages converted from text-based PDFs

This enhancement was published in version 3.1.31.68 (2025-03-27).

ENHANCEMENT #2061: REMOVED INTERNET EXPLORER DEPENDENCY: We have removed the obsolete IE dependency used for Oauth20 login boxes such as for Import, Forward and Export Email, Export to Box etc. and we have replaced it with WebView2 (Edge).

This enhancement was published in version 3.1.31.67 (2025-03-18).

ENHANCEMENT #2054: VALIDATE – DB LOOKUP: The “Confirm Lookup” option now only asks for confirmation if the user changes the value.

Before, the “Confirm lookup” option forced the use to hit ENTER twice even if the user did not change the value.

The double ENTER is now only required if the user changes the value. With the idea that the first time the user hits ENTER the lookup is triggered and all fields are mapped. Then the user checks and confirms the lookup with the second enter.

This enhancement was published in version 3.1.31.67 (2025-03-18).

ENHANCEMENT #2062: SERVER – MONITORING: We also added CPU utilization to the memory consumption monitoring file.

This enhancement was published in version 3.1.31.67 (2025-03-18).

ENHANCEMENT #2072: LOGGING: We now log when MetaServer is started. The log is, as usual, located in:

C:\CaptureBites\MsData\Log

This enhancement was published in version 3.1.31.68 (2025-03-27).

Version 3.1(30) | 2025-02-11

NEW #1969: EXPORT TO SHAREPOINT – DYNAMIC SUBSITE SWITCHING: In the Export to SharePoint action, it is now possible to use variables to set the SharePoint address. Each SharePoint subsite should use identical credentials, libraries and content types.

NEW #1989 #1998: EXPORT TO WEB SERVER – ZAAKSYSTEEM.NL: We can now export to zaaksysteem.nl. Related to this we also added a check in Validation to check the validity of a “zaaknummer” through a web call.

We also don’t stop queue if the document upload error is one of the following:

“Zaaknummer bestaat niet”

“Ongeldig zaaknummer”

“Upload missing.”

Such documents are moved to the Errors tab. Other (correct) documents continue to be exported.

NEW #2001 #2007: EXTRACT – SET FIELD VALUE: We added new variables:

1. In the Import / File section, we added { Import File SHA256 }. It returns the SHA256 string of the imported document.
2. In the Document section, we added { Processed PDF SHA256 }. It returns the SHA256 string of the PDF(s) after manipulation in the Separate Document / Page Processing action and / or Organizer.

NEW #1995: EXTRACT (AZURE AI DOCUMENT INTELLIGENCE): We added the latest GA version 2024-11-30 (4.0 GA) which is available in all regions.

We recommend to switch your existing Extract (Azure AI Document Intelligence) rules to this GA version. Any newly added rules automatically set the new GA version.

There are extraction improvements in all models and new fields are available in the prebuilt Receipt model:

• Balance
• Balance Currency Code
• Balance Currency Symbol
• Country Region
• Payment Item Amount
• Payment Item Amount Currency Code
• Payment Item Amount Currency Symbol
• Payment Item Method
• Receipt Type
• Tax Detail Item Description
• Tax Detail Item Net Amount
• Tax Detail Item Net Amount Currency Code
• Tax Detail Item Net Amount Currency Symbol
• Tax Detail Item Rate

ENHANCEMENT #1899: VALIDATE – DB LOOKUP: Selecting the first value in the lookup list on a non-required lookup field with the cursor required going up and down with the cursor. Also, selecting a blank value was unintuitive.

The new improved behaviour shows a blank first line when opening the list, selecting the blank line, keeps the field blank. If you start typing, the lookup shows all matching values and no blank line. If you want to select the first selected line, you just have to hit ENTER once.

FIX #1942: EMAIL FUNCTIONS – OFFICE 365: We now retrieve proxy settings from the registry:

HKEY_CURRENT_USER\SOFTWARE\Microsoft\Windows\CurrentVersion\Internet Settings to set the proxy settings correctly for all email communication.

FIX #1973A: PREPROCESSING LARGE LZW FILES: Preprocessing failed on very large TIF LZW files. We fixed this by handling this part of the processing with a 64bit executable.

FIX #1973B: CONVERT TO TIF: When converting electronic PDFs with very large logo bitmaps (we had a case with 2187 dpi logo bitmaps), they were rendered to TIF with the same resolution as the embedded image causing memory issues. We now limit the resolution to 400 DPI. We applied the same fix to the following actions:

– Convert to Image PDF

– Convert to JPG

– Digital Imprinter

FIX #1966: ERROR REPORTING – ALL AZURE INTEGRATIONS: We now display the inner error giving more meaningful information about what causes the error.

FIX #1967: EXTRACT BARCODE: The folllowing error was caused by a defect in the retry logic:

“Unable to deserialize MS.Common.Exceptions.ExternalExeException”

FIX #1975: EXTRACT – FORMAT CSV: The “Minimum number of rows” spin box was not correctly anchored when resizing the window.

FIX #1957:  ERRATIC ERRORS: Unable to delete file… errors started to appear since version 3.1.28 related to following actions:

– Check Validity

– Delayed Apply

– Distribute

We changed the polling delay from 50ms to 5000ms for these actions to check if this may fix the issue.

FIX #2002: EXTRACT – AZURE AI DOCUMENT INTELLIGENCE: Models returning the dominant language and document type always returned a confidence level of 0% for these values.

FIX #1965: VALIDATE – DB LOOKUP (METASERVER DB): When you disabled the “Check if multiple hits option”  and the DB contained equal lookup values in both upper and lower case. then the document appeared unexpectedly in Validation and also returned the wrong lookup result.

FIX #1977: EXTRACT / VALIDATE – DB LOOKUP (METASERVER DB): When you added a field or removed a field from the CSV, then accessed the field list via a DB Lookup setup, the change was not reflected. It required to restart MetaServer to see the changes.

NOTE: If you are already in the DB Lookup setup and you make a change to the CSV, press “Connect” in the setup to load the new version of the database.

FIX #1970 #1981: EXTRACT / VALIDATE – DB LOOKUP (ODBC ORACLE): We experienced 2 erratic errors related to lookup with Oracle ODBC:

“Unable to identify the syntax required by this database”

“La connexion a été désactivée.”

We now retry for the following errors (both SQL and ODBC); 4 retries, after 0.5 sec, 1 sec, 2 sec…:

“Unable to identify the syntax required by this database.”

“Unable to identify the case syntax required by this database.”

In case of “La connexion a été désactivée.”, when an action connects to an odbc and an existing connection is used, a call to GetFields is made to check the connection. If this call fails, a new connection is made.

FIX #1951: AZURE AI SUMMARIZE: We have made adjustments to the settings for abstractive and extractive summary.

Extractive summary features the option: “Sentences in extractive summary”, a value between 1 and 20 (default = 4)

Abstractive summary features another option with 4 possible levels (default = short):

– One sentence of around 80 tokens

– Short of mostly 2-3 sentences, with around 120 tokens

– Medium of mostly 4-6 sentences, with around 170 tokens

– Long of mostly over 7 sentences, with around 210 tokens

FIX #1962: AZURE AI SUMMARIZE: The info tips were not correct. They are now:

“Apply” info tip:

“Only summarize if the condition applies.”

“Summary type” info tip:

“An abstractive summary produces a summary with novel sentences. An extractive summary produces a summary using key sentences from the source text.”

FIX #1912: SEPARATE DOCUMENT / PROCESS PAGE: Separating every n pages without any extraction rules caused an error when previewing. Run time worked fine.

FIX #1997: SEPARATE DOCUMENT / PROCESS PAGE(S): Setting a field equal to { Document Page Number } or { Document Page Count } during Separation if multiple documents in a folder are merged during import caused an error.

Another related thing was that if you used the { Document Page Number } variable directly in the condition of the “Extract Barcode” or  an “Extract Text” rule, it did not resolve.

FIX #1945: ORGANIZER – CLIPPING TOOL: A2 PDF files generated correct thumbnails, but the wrong clips.

Also, generating several clips from the A2 caused an OOM. The images were buffered to avoid rasterizing the page again when previewing a page a second, third… time. We removed this buffering to avoid the OOM issue. We also moved the rasterization in an external exe.

FIX #1976: VALIDATE – DB LOOKUP: In the DB Lookup’s setup, on small screens with a lot of fields in the mapping tab, you could not get to the bottom fields to map them because the vertical scroll bar was missing.

FIX #1960: VALIDATE CSV: The “Height” expressed in number of lines in the setup did not correspond with the exact number of lines shown in the CSV table. For example, when you entered 10 lines, only 9 were shown.

FIX #2008: EXPORT TO FOLDER – XML FILE INDEX: Any MetaServer fields included in the XML header are now only updated by the first imported document. Consecutive documents leave the MetaServer field values in the XML header untouched.

FIX #1992: SERVER TAB – ACTIVE QUEUES ONLY VIEW: An automatic or manual refresh unselected the selected action.

FIX #1968: LICENSING – AZURE AI SUMMARIZE: Licensing now works correctly. You can always test your Azure AI Summarize rule during setup. However, during runtime, if the module is not licensed, documents passing through an unlicensed Azure AI Summarize rule will return the following error:

“Module ‘AzureSummarize’ is not licensed.”

FIX #1968: LICENSING – AZURE AI SUMMARIZE: Licensing now works correctly. You can always test your Azure AI Summarize rule during setup, but, during runtime, if it is not licensed, documents passing through an unlicensed Azure AI Summarize rule will return the following error:

“Module ‘AzureSummarize’ is not licensed.”

FIX #1980: LICENSE TAB: We have removed the “VRS QC Client” module. This module was never released and is not on the roadmap anymore.