Instabase FAQs

New Features

Application

Diff app now contains an audit log as well as various metrics tracking

Audit log contains usernames as well as filenames the Diff App has been run against.

Metrics increments on number of calls and by username, diff mode, extensions.

File metadata is also logged.

Redactor:

Redaction boxes can now be positioned relative to others.

Redactor can now redact results found by Refiner Functions.

A new Run Step API has been added to the Flow API.

Added Global Address Token Matchers!

These allow Token Matching for for entities such as cities, states, countries, zip codes, and local address classifiers.

Added a Newsapi.org Fetcher, which is included in new Adverse News projects.

Added a new Refiner Function called is_ocr_required which determines whether a record required OCR for processing.

Platform

Microsoft SQL Server is now supported as a back end database.

Changes

Application

The 'Use New Uploader" option when uploading files has been removed as the "New Uploader" is now the default.

A more easily accessible way to execute UDFs is available in Text Editor for those users with elevated privileges who can edit and execute these kind of functions.

Provenance tracking now supports multi-page documents.

Site Administrators can now update organization ownership.

Launcher and UI has been updated with the new Instabase logo and theme.

Bug fixes

Fixed issue where havingSITE_SETTING_ONLY_SITE_ADMIN_CAN_EDIT_ORG set could, under specific circumstances, cause users to not be able to view organizations they were a member of.

Fixed issue where Diff App would incorrectly ref the right corner in some edge cases.

Fixed two exceptions in Visual Refiner

When a template and input documents had different numbers of pages.

When a page contains no annotations, followed by a page with annotations.

Visual Refiner now has a configurable timeout.

Fixed a rare bug with HQ Document OCR that was producing inconsistent results.

Redactor Generate Replacement now creates tokens with roughly the same length as the original token.

Documentation

Diff App documentation has been added.

New Features

Application

Upgrade NLTK to version 3.4.5.

Add mapped records to the Metrics API.

New utilities for writing custom fetchers, such as HTML parsing, request retries, and common search terms for adverse news.

Added ability to monitor news streams for news about a particular company's acquisitions

Introduces an Apply NLP Flow step for various natural language tasks such as sentiment analysis and entity-action analysis

Platform

Add support for deploying using the PostgreSQL database.

Instabase now allows using Hitachi Vintara's S3 platform as the file-system.

Documentation

Adds two Classifier guides, How to Use Classifier and Custom Classifiers, to the docs.

Changes

Application

The App Settings page has been removed as it's been deprecated.

Visual Refiner is now compatible with Flow V2.

Added provenance tracking to regex_find and regex_get_all.

A new interface for Tokenizers has been introduced which support provenance tracking.

Tensorflow has been updated from 1.5.0-rc1 to 1.14.0 to address a security vulnerability.

Redactor will not align documents if there are no absolute positioned annotations.

Default to annotate first file with Redactor and Visual Refiner

Platform

Bug fixes

Fixed two Redactor exceptions

Exception occurring when text is too large for a given region.

Custom text overlays not working with absolute annotations.

Fixed exception caused when Process Image Path is invalid in Review OCR

Fixed issue causing Classifier V1 to fail on occasion.

Fixed issue where signature detection would start even when signature image dimensions were zero.

The ibocr builder was being reused, which meant that classes were not actually being filtered from the output. Start with a new ibocr builder so that we can properly filter the classes.

Fixed bounding box issue that resulted in some words being dropped.

New Features

Application

In the Flow Editor view, a user can choose to run a Flow step with Settings using the Run As option.

Redactor supports redaction with black/white boxes or blur, and overlaying custom or auto-generated text

ACL permissions are now enforced when using Annotator-based apps.

Custom TokenMatcher and Tokenizer registration within UDFs has been added. See documentation for more information.

Signature detection accuracy improved.

OCR now supports barcodes.

Platform

In the Grant Access page, a user with the site:manage_teams ACL permission my use the new Teamstab to manage permissions for all teams across all organizations.

A new panel within the Grant Access page called Site ACLsenables site admins to set the Manage Users and Manage Teams site-wide ACL permission. This permission can be set independent of that user needing to be a site admin.

File system mount details can now be encrypted via API.

Changes

Application

Diff app now has a wrap text option to wrap diffs that don't have line breaks.

When running a Metaflow (v1), additional ibocr files will not be generated directly in the out/ folder, but rather in the out/s1_process_files folder. This results in the more expected behavior of ibocr file locations.

Annotator apps now default document zoom by height of the tallest document, not width. Buttons to adjust the zoom have been added.

Changes to retry options for Flow execution in batch mode:

Retry delay: 180 sec -> 30 sec

Max retries: 3 -> 2

classes.json and related output are now always created when running a Binary Flow.

Redactor, when run in a compiled Flow program, will output individual images as well as the stitched together PDF.

The directory containing results from a Flow run is now pinned for easier navigation to the results.

Platform

Defining a registration function within a UDF is no longer required.

Increase the default timeout of Classifier step to be 30 mins

Bug fixes

Fixed an issue where the status update in batch mode Flow N/0 files was displayed instead of the the actual number of files.

Fixed a bug in Redactor UI outputting the same input file.

Fixed permissions in Launcher UI that erroneously allowed everyone to see options to mount and edit options for drives and databases

Though the links were erroneously shown, they did not function unless the user had proper permissions.

Fixed an issue in that would cause the final Flow of Multi-flow execution to fail to run.

Fixed occasional broken Breadcrumb links when using the Recipe Book app.

Fixed the review OCR pages ability to account for corrections to word orientation.

Provenance tracking now displayed correction in review OCR when pages have been rotated.

Security

Both thescikit-image andNumPyPython libraries have been updated to versions 0.14.4 and 1.16.5, respectively. These upgrades addressCVE-2019-6446.

Any User Defined Functions using those libraries (containingimport numpy) should be checked to confirm they function as expected.

Documentation

Fetcher documentation update to show how to use IBFile module within Custom fetchers.

Flow documentation now include how to add anApply Redactor step.

Redactor instructions needed for setting up a teamplet directory added.

TokenMatcher for Names documentation added.

Tokenizers and TokenMatchers for US Addresses documentation added

Updated Metrics API Documentation with Flow Results Metrics Information.

New Features

Application

Added the ability to use CMD/CTRL+S to save program within Refiner and Refiner V2.

Added export API to diff app that supports formats like PDF and HTML.

Added new Marx OCR Model option to Process Files.

Added Instabase ACL support to Redactor. A user can only access the app if they have can_read, can_write, and can_executepermissions.

Added support for multi-part LDAP authentication.

Addednltk.tokenize.punktsupport for natural language token matching.

The Flow Binary runtime config is now accessible in the custom FetcherContext via the runtime_config property.

The Flow Binary runtime config to now accessible in Apply UDF via the CONFIG variable.

Added a new set of Refiner functions dedicated to handling maps.

For example,{'key1': 'val1'}

Platform

New Flow Resultsmetrics are available via the Metrics API.

The Metrics API is now able to group results by time interval, and we updated the documentation to match.

Changes

Application

Various improvements to Metaflow UI and button text.

The Diff app supports changing between word by word andline by line algorithms.

Annotator apps (Annotator, Redactor, and Visual Refiner)

The modals for Creating, Exporting, Opening, and Running programs attempt to infer which file paths to use based on what is open, to reduce the amount of clicking in file pickers.

Show all pages in a single view, rather than having to paginate through them.

Classifier will now look and behave more like the newer Instabase apps. Instead of viewing and editing Classifiers from a list view, open individual .ibclassifier files to see details, train them, and edit them. This only affects the UI of Classifier, not the functionality. See the updated documentation for details.

Platform

Flow Steps involving UDFs (User Defined Functions) & Refiner App require the programs to execute within 600 seconds for a single file / record. The following Flow Steps are affected:

Apply Classifier

Apply Map Records

Apply Refiner

Apply UDF

Apply Sheet

Apply Visual Refiner

Apply Redactor

When Flow is run in binary mode, the timeouts will cause the individual records to fail, whereas when run without binary mode will cause the entire Flow to fail.

Fetch File flow step now has a time limit of 1800 secs.

Remove logic that allowed switching back to the old navigation UI

In apps that have a toolbar at the top, the toolbar is now fixed to the top of the screen even when scrolling.

Bug fixes

Fixed Redactor's poor handling of multipage documents.

Fixed issues with filenames containing special characters.

Fixed a bug wherethe link of the last breadcrumb was incorrect in the file system.

Fixed the bug where pdfs with pages of different sizes were not working with provenance tracking.

Fixed a bug in redactor where multi-page documents would not be properly processed if any of the pages had no annotations.

Fixed bug downloading articles for NLP Search.

New Features

Application

The Flow Binary runtime config is accessible in the custom Fetcher context via the runtime_config property.

A variable called ENV, which is a map<string, string>, isaccessible in Refiner and can be passed into a running Flow Binary via the API.

ALPHA:Introduced a new framework for building and running NLP pipelines. This framework is experimental and the corresponding API is subject to change.

ALPHA: Diff App allows users to compare two versions of a file, save the comparison and re-open it.

Within a Refiner and Sheet custom function, a user may access an ibfile object that can performread_file and write_file operations with the permissions of whomever is running the operation. This ibfile is accessible via the function context passed in through the kwargs to the custom function.

Custom Fetchers can now access the ibfile object using the FetcherContext, which allows for file access using the interface for needs like reading and writing files. This ibfile object will be authenticated based on whichever user executed the Flow.

Platform

Translation support: Introducing the parameter translation in Process Files. This will help to specify if the content needs to be translated into a specific language. For example to convert from Vietnamese to English the parameter will look like this: "translation": "vi:en". In case of unknown source language the parameter will be: "translation": "auto:en".

Infrastructure

Instabase can now run on a dedicated set of nodes in a Kubernetes or OpenShift cluster.

Added support for automatic Index Lifecycle Management in Elasticsearch (configurable, periodic index data compaction and deletion).

Changes

Application

PDF functions in Refiner accept INPUT_IBOCR_RECORDas a parameter. This is the new recommended way of using these functions instead ofINPUT_IBOCR as a parameter. Use of INPUT_IBOCR will be deprecated in September.

The map_records custom functions: ibocr_scan_repeated, match, and splitwill return and error if the input parameter is not ParsedIBOCR.

When running Flow through compiled binary, the input files are now processed in batches instead of being processed all at the same time.

This behavior can be overwritten in the UI by un-checking the option Run Flow in Batch Mode in the Extra Settings section.

This behavior can be overwritten in the API by setting the parameter use_flow_batch_mode: truein the settings section.

The max batch size is 40.

The batch size can be changed in the UI by setting the Batch Size field inside the Extra Settings' section.

The batch size can be changed in the API by setting the parameter 'flow_batch_size' in the 'settings' section

Running Flow binaries now have a different interface. Pick an input folder first, then choose additional settings, finally there's an option to upload pasted text to the input folder first before running the flow on the input folder.

Map records by default will not strip the metadata list in Visual Refiner. Since there may cases where the resulting output files would be too large, we have also added a flag that will allow the user to choose whether to strip metadata_list from the ibocr or not.

Emerson may now be run across multiple document types (IDs and longer-form documents such as letters/forms), and auto-detect doc-type if desired. Default is still set to IDs.

Platform

An environment variable toggle whether or not site admins are granted Admin access to all Organizations across Instabase, even if they're not explicitly in the organization.

In the file system, sometimes the user would see actions (like "rename" or "delete") that they didn't actually have access to based on their permissions. They wouldn't be able to successfully make these actions, but they shouldn't see the options in the first place. This change ensures that a user only sees the actions they're actually able to make.

Bug fixes

Fixed the maximum cell limit for XL sheet. If the input string is > 32672, we truncate it at 32672 characters. This matches the cell width limit in Microsoft Excel.

Fixes web server startup procedures to ensure web servers don't crash when RabbitMQ is unavailable.

Fixes a bug in Launcher that caused the filter bar to disappear if searching for a term that has no results in the organizations view.

New Features

Application

Enable Refiner Functionselection in Redactor.

Add new refiner logical function IF_ERRORwhich behaves like a try...catch block.

IF_ERROR( expression, if_error_expression )

If expression does not throw an exception, return expression.

If expressiondoes throw an exception, return if_error_exception.

If both throw exceptions, throw an exception to the refiner program writer as a usual refiner ERROR.

Refiner v2 now supports the ability to add and remove records/example inputs to test your refiner formulas on. Right click the Rowand select the Insert Rowor Delete Rowoptions.

Flow now supports combining multiple Metaflow binaries that use Auto Classifier into a single binary. This enables users to add new Flows to an existing Flow binary.

Platform

Access token applications within the Developer SDK can now be deleted. Access tokens corresponding to deleted applications are also invalidated.

Added support for automatic Index Lifecycle Management (configurable, periodic index data compaction and deletion).

Added the ability to extract an uploaded zip file to destination folder via Filesystem API. Please note there is a 100M filesize limit.

Changes

Application

Provenance tracking highlighting is improved, as well as adding automatic scrolling to tracked fields within Review OCR.

Added option to Clear Metadata when running during Map Records, which can create smaller output files. See Map Records documentation for more information.

The Metrics and AuditLog APIs now support "pretty" indented JSON responses, by adding ?pretty=true the end of the query URL.

Bug fixes

Flow Run UI Status correctly updates to Success when the Flow completed successfully in the backend.

Process files run OCR correctly on JPEG compressed TIFF files.

When clicking the results banner in the UI, we now take you to the step's output directory like you expect, rather than to the root output directory.

In Review OCR the icon popovers on field labels now disappear when you click outside of them, instead of staying on screen until you click the icon again.

In Refiner V2 cells correctly change from red to normal when when all of the errors in the row are resolved.

Documentation

Support for extracting zip files added to Filesystem API.

New Features

Application

NLP Search

New "real-time" news searching interface with simpler UI and extensible search & analysis profiles.

Custom profiles for NLP Search may be edited in editor for easy creation and updating of existing NLP Search Profiles.

Improved news article search & download speed.

Added ability to redact by token matching and replace with fake data.

Users may now use Refiner V2.

Refiner V2 is similar to Refiner Classic but with two main differences:

Refiner V2 only shows one record/row at a time: user can toggle between which record/row to display.

Refiner V2 does not support AUTOEXTRACT() formula.

Refiner V2 has much better support for large Refiner files and for operations like insert and delete column. Since Refiner V2 shows only one record/row at a time, it also has a Run settingssetting to only load results for a single record/row at a time vs load results for all records/rows.

AddedAutofind as a position option in Redactor. Fields that have the Autofindposition will show in the interface where the user drew them, but when documents are redacted they will perform a search against the associated token matcher and redact all of those instead. For example, you could select to Autofind Social Security numbers and, as a result, redact all Social Security numbers at runtime.

Changes

Application

In Refiner V2 you can now navigate to the previous/next record with errors quickly using two red buttons at the top-right of the page. These two red buttons appear when there is at least one record with errors.

Platform

Nested folders can be created by specifying the path when creating a new folder in the file system.

Bug fixes

Classifier scripts were omitted when binarizing a classifier step. We now support binarization of Classifier steps with custom scripts (e.g., Custom Classifiers).

Fix the copy link button in file system menus so that it gets the correct link instead of one with false in the middle of it.

On the Flow Review page (after opening .ibflowresults), you can now click View All Recordsto visualize all records.

Refiner formula versioning works when opening a new .ibprog from a directory of files now.

Fixed occasional errors where Flows were misconfigured resulting in exceptions with messages like NoneType specified.

Documentation

Updated documentation for Apply UDF Options file client.

Updated Refiner documentation for allnlp_token_* functions.

New Features

Application

Major update to Recipe Book which adds setup wizards and built-in recipes.

NLP Search now has a "realtime search" option that performs a search in memory without any project setup.

A user may generate backup codes when TOTP or SMS is setup for multi-factor authentication.

For new ibprog files, Refiner will by default use the most recent formula version.

Classifier can now be run as a step.

Refiner

Refiner will now show a save status:

After a change is made => "Unsaved changes" will show.

After the ibprog is saved to a program => "All changes saved" will show.

Support for changing Refiner formula version through "Edit Program Settings" modal.

Starting from Refiner formula version 2.1, formulas support wider use of Unicode (i.e. better support for Chinese characters).

Platform

In the "About" section, we now provide an "About Instabase" tile which links to version information about your Instabase deployment.

Infrastructure

Debug logging for the time needed to run many file system operations as well improvederror logging.

Changes

Application

Fixes the bug where we cannot use integers in refiner programs. Now, if the numbers have a "." in it, they will be parsed as a float; otherwise, they'll be parsed as an int.

Previously, access to ACL manager for organization-owned repositories was limited to site administrators. Access has now been added for organization administrators.

Metrics and AuditLog APIs properly differentiate between username and email when appropriate.

Platform

Automatically expand folders after creating a new file or folder in them in the file system.

Infrastructure

Upgraded Instabase NGINX from version 1.15.9 to 1.15.12.

Bug fixes

Fixed a bug that was preventing Nginx from properly routing Recipe Book API requests.

Fixed a bug when compile binary returns NoneType.

Fixes a bug where Process Files was failing when given a JPG image in Palette mode. (This most likely happens when a PNG in Palette mode is saved as a JPG and is then passed into the system).

A Flow web-hook request now adds a Content-TypeHTTP header set to application/json to indicate JSON content in the request body.

Documentation

Added alpha functions filter, map, and flatten_listas advanced ways of manipulating lists. We encourage experimental use and feedback but would caution against putting in production.

Added a warning clause in documentation informing users of the dangers of using cur_status field in Status API. Status API is used for polling status of a Flow.

New Features

Application

Report confidence and provenance fields in Visual Refiner IBOCR output

When the launcher view is enabled, a user can now choose between a grid and a list view for Workspaces, Drives, Databases, and Apps

There is a new flag for high quality scanned documents that will try to account for tilts in documents. See reorient_words underOCR config in the documentation.

Documentation for AuditLog API added

New OCR config in Process files called extract_all_pdf_layers. This flag will run OCR on all PDF pages, as well as extract the machine-readable text elements, and stitch it together into one page of text. This operation is expensive but gives you the highest quality results for machine-readable and PDF pages with images embedded in them

Allow users to set confidence threshold for checkbox decoding.

Audit Log API added.

Show annotation type icon in Visual Refiner sidebar.

If youre using Launcher, you have the option toggle between a list view and a grid icon view for each of the views.

If youre using Launcher, easily navigate to Instabase Support by following Get in touch in the Help section.

Option to include the actual image processed with any Visual Refiner field.

Documentation for Metrics API.

Team collaborators can now be modified and deleted via the API.

Platform

Organization level management of default repository ACLs is now possible via API.

Users can manage default repository folders and default repository ACLs either in batch or individually.

UI Launcher can now be enabled and disabled via environment variables.

Updating and deleting teams and managing team members is now available via API.

Updating and deleting organizations and managing members is now available via API.

Changes

Application

Under organization settings > Default folders tab > the contents of the text box are now populated with a JSON object instead of a list of folders.

Improved Text Extraction with PDF documents.

New look for the file system that will not impact functionality.

Improved checkbox decoding module.

Metrics API can now be used by non admin users to query their own metrics.

All Flow Binaries will run in batch mode by default.

Platform

Improvements to HTTPS enforcement.

Bug fixes

Fix a bug where some large corrupt images files would fail during OCR.

Bug fix to allow users to specify a page range in Process Files.

Fixed some error messages that displayed poorly when an error occurred using Flow.

Bug fixes with NLP search.

Fix recipe book status reporting.

New Features

Application

Visual Refiner's IBOCR output now includes provenance tracking and confidence reporting.

Recipe Book

Major app update including:

New UI.

Premade Recipes from Instabase.

Installer wizard for easier project setup.

Integration with Flow's Run view.

Project-specific recipe books for project management.

Screen reader support enabled.

If an error occurs while running a Flow Binary, Instabase will link to the IBFLOW.

You can now toggle between list and grid view in Launcher.

Platform

You can now get and set SAML Mappings via the API. Please refer to the documentation for more information.

Changes

Application

Markdown (.md) files are now supported and rendered.

Metrics and AuditLog APIs properly differentiate between username and email when appropriate.

Launcher menus now open on click instead of hover.

The sidebar's visibility can now be toggled.

Not being logged in now presents a minimal sign in page instead of including Launcher.

Platform

All services now support logging levels.

SAML Mappings integrate independent of organization the isolation flag.

Documentation

Significant Additions

Warning clause in added informing users of the dangers of using cur_status field in Status API. Status API is used for polling status of a Flow.

AuditLog API documentation for each audit log type.

Site API documentation.

General Launcher documentation.

SAML Mappings documentation improvements.

Bug Fixes

Application

Fixed issues with mounting databases in Launcher for some users.

Allow users with read-only permissions for a folder to copy it.

Fixed issue where applying organization defaults occasionally did not work.

Saving default workspace ACLs in organization settings now refresh the text editor upon success.

Corrected an issue with correct_orientation_fine when converting to greyscale. This flag corrects skew as a preprocessing step, allowing for better OCR.

Fixed an error that occurred when a word is outside the bounds of the image and thus it's bounds are undefined.

Main navigation bar has had it's Z-Index adjusted and fixed.

Fixed some possible scenarios where default ACL verification would fail.

Minor fixes to how the Beta Apps section displayed.

Added corrected CSS library to Table Refiner inside of Launcher.

Opening links in File System will now open in a new browser tab, when appropriate.

Breadcrumb trail links now properly wrap.

Platform

Fixed Metrics API issue whereend time was being set to start of the current day instead of the current time.

Metrics API nowproperly match counters.

Metrics API now uses a correct start value.

The Run Binary API webhook now sets the correctcontent-type.

New Features

New Launcher UI

Customer may enable the Launcher UI using the use_launcher=true cookie. Please contact Support for more information.

Refiner

When training a Refiner program, a user may specify a "row range" to speed up the training process and getting results back faster

Refiner can now support the following sequence of steps: Process Files -> Visual Refiner -> Refiner.

The result will be the combination of Visual Refiner & Refiner output. Please note that Map Records and Refiner -> Visual Refiner are not yet supported.

Three new Refiner functions added: map, filter, and flatten_list. These facilitate using lambdas with lists.

OCR

Enable the OCR Emerson service for specific customers

Option to embed an image of certain refined fields in the output. User can view these images in ReviewOCR and data is available Base64 encoded in the IBOCR.

Review OCR and Review Batch page support rendering of images when a column starts with the "img_" prefix

Confidence flagging within Review OCR can now be configured within Refiner Settings (available when confidence tracking is enabled)

API

An administrator API has been added to view various metrics. Please see documentation for details.

Repositories can now be created, deleted, and viewed via the API. Please see documentation for details.

Organizations, Teams, and Collaborators can now be created, deleted, and viewed via the API. Please see documentation for details.

Documentation for the Metrics API added

General

File -> Export functionality added

Intended primarily for Instabase use initially. Please let us know what export formats are useful to you and we may be able to add.

First version of Metrics application is available for use

Added context links for copying links, paths, and names of files and folders

Files with JSON content are now pretty-printed when the user chooses to "Preview" the file

Infrastructure

Added support for HTTPS only mode within Instabase webservers, support for customer proxies

Customer can use an environment variable in the yml whether to enable the new Launcher user interface or not

Added support for AWS Elastic Load Balancer

Changes

Application

Refiner programs will no longer run automatically and Run must be clicked

Re-design of Refiner's menus to use our standard IBApp format

Refiner can now be run on a subset of rows through Refiner Settings for quicker debugging of Refiner results

Hidden fields can now have their visibility toggled within Review OCR

Provenance tracking can now be tracked automatically for simple extractions (confidence tracking and auto provenance must be enabled)

Render any README.md in directories that have this file as a child

Change to the menu look-and-feel, using the IBApp common template

Visual Refiner's URL is changing from visual_refiner to visual-refiner to conform with other Instabase applications

Infrastructure

Openshift only: NGINX now does not use DNS resolver and relies on ClusterIP for each service running

Bug fixes

General improvements to page-load time

General improvements and bug fixes to confidence tracking

General performance and reliability improvements of Flow Batch mode

Fixed incorrect nesting of directories when coping or moving files across repositories

Improved highlighting of tracked results within Review OCR

Fixed reporting of status = ERROR in the Flow Dashboard app

Various OCR bug fixes for Map Records

Fixed an issue with Fetch Files step when no files were being produced

Fixed a UI bug where tree picker sometimes rendered incorrectly

Fixed a UI bug where sometimes file paths could be interperated incorrectly

Updated base Docker image to for security updates to various processes

Fixed invalid JWT authentication errors related to incorrect cryptodome type

Now using Ubuntu 18.04 based image for Nginx to avoid vulnerabilities related to previous Debian image.

Fixes to Table Refiner to improve scrolling behavior and enable specifying error tolerance

Fixed the "Preview" option in the File browser, such that different files can now be opened with the "Preview" link

Fixed NLP Search issues when fetching webpages

Improving messages within Table Refiner

Infrastructure changes

When a worker is lost, we retry the task

Impact: We expect an improvement of total number of Flows which complete, even in the event when an infrastructure level change occurs (like a pod dying).

PDF page results for a given file are now cached for 24 hours to prevent re-computation.

We internally split PDFs into multiple pages and then re-join them to reduce memory usage of PDFService.

Each step in a Flow is now limited to 2 minutes at maximum.

Each image produced per PDF page is limited to 30MB

Reduce disk memory usage for files emitted during a Process File run

Improve quality of JPEG processing

App changes

Process Files:

produce_metadata_list is now to set false by default

Impact: The height / width of documentswon'tbe returned in the final IBOCR

produce_word_metadata is now set to false by default

Impact: The coordinate level metadata for extracted words will no longer be available in the IBOCR

A newer OCR model is used by default

Impact: We expect that overall, we will get better OCR results with this newer deployed model. Lets monitor closely how many successful times we OCR a page.

Preview/Experimental Release: Emerson OCR: Targeted at poor-quality ID cards. Currently for experimentation on 1-3 documents at a time.

A new fonts flag has been added to OCR options, which will help OCRing specific fonts like OCR-A, OCR-B, and MICR.

A correct_resolution_auto flag, which should is an improved version of resolution correction for images

UI changes

Roll out new file browser UI

Bug fixes

Fixed regression where users using JWT authentication received an HTTP 400 authentication error.

Standard Support offers English language support during the hours of9am to 6pm Pacific time (GMT-7) for all priorities except where outlined in your Instabase license.

Additionally, Support is closed on the following United States holidays:

New Year's Day

Martian Luther King Day

Presidents Day

Good Friday

Memorial Day

July 4th

Labor Day

Thanksgiving Day

Christmas

Your license may include a specific SLA for various ticket priorities. Please see our priorities definition for more.

For more information about your support contract, please contact [email protected].

When submitting a new support ticket you'll need to select one of four priorities for your request. Please use this guide to set the appropriate priority.

Low - Information requests, feature requests, general questions and assistance.

Normal - Configuration help, non-critical bugs, non-production system failures.

High - Production malfunctions impacting business systems. This can include performance issues or non-critical bugs.

Urgent - Fatal production system failures impacting critical business systems.

Instabase Support may change the priority of your ticket if the selected priority does not meet the above criteria or if the priority of the ticket changes through troubleshooting. For example, if an urgent ticket is created and the urgent aspect of the ticket is dealt with, but follow up analysis is needed, the priority of that analysis may move from Urgent priority to High priority.

Instabase offers English language support for Instabase on-premises, virtual private cloud, and Instabase.com customers. You can create a support ticket by visiting our Support Portal or by simply emailing [email protected].

While not required, we encourage users to create a support account and use the Support Portal as it allows for easier tracking and prioritization of requests. It can also helpexpedite your request.

Instabase Support is here to help in a wide variety of situations.If you're uncertain that Instabase Support is the right place to ask for help, always feel free to open a support ticket and we will work with you to determine the right place to get help.

However, there are some situations when Instabase Support is not the right place for help:

You are working with Instabase on an deployment and already have an Instabase technical point of contact.

You have questions on a tool that is unrelated to Instabase. Please use support channels for that tool.

You have a sales question. Instabase's sales team can be reached at [email protected].

You require significant custom scripting (i.e. User Defined Functions, etc...). Please send a message to [email protected] for information on custom services Instabase can provide.