Release Notes

UI Refresh – All Pages

We’ve refreshed our UI with new fonts, colours, and styling.

Models – Add training data to custom model through the front end

Previously, when using custom models, the only way to add new training data was through the API using the parameter [outputToMaster=True]. Now, when enriching with a custom model in the Studio, a toggle will control whether the enriched data is added to future training data. If the “Use enriched data for future training” toggle is set, your enriched data will be added to the model’s master enriched dataset. From there, corrections to the training data can be made via API or in the Models screen.

Optimizations & Security Improvements

We’ve been continuing our work on our SOC2 type 2 certification.

Stage – Saved Report Reusability

Previously, saved reports were only available to be applied to the original dataset used to create the report, and to datasets resulting from enrichment jobs of that original dataset. Now, you can reuse saved reports on any dataset within the same Project, so long as the relevant columns are still present

Enrichers – Sentiment Analysis

LLM Sentiment Analysis has been renamed Sentiment Analysis. This enricher uses your environment’s LLM to output three columns: A sentiment label, an emotion label, and sincere/sarcastic label.

The previous Sentiment Analysis enricher has been renamed Sentiment Analysis (Offline). This enricher runs offline, and outputs a classification (Positive, Negative, Neutral) and a confidence score.

Jobs Screen – Usernames on Jobs Screen

In the Jobs screen, the username of the user who started an enrichment job will be displayed. This will help you understand who is running enrichment jobs in your environment.

Jobs Screen – Time Display

When you select a completed, cancelled, or failed Job in the Jobs screen, the Inspector displays the duration of the Job in a more readable format now.

More LLM Options

We’ve added more LLM options to use in your environment. Please contact support if you would like to change your environment’s LLM.

Bug Fixes

• File Selector – Duplicates in Dropdown: Sometimes, the file selection dropdown in the Stage and Studio showed many files in duplicate. This has been resolved.

Line Graph for Persona Segmentation

When you run the Persona Segmentation enricher and select Automatic Clusters, the platform uses an algorithm to determine the optimal number of clusters. Now, the data representing this number is available in a table, which can be displayed as a line graph in the Stage.

Please note if you select to choose a custom number of clusters, this file and chart will not be created.

LLM Sentiment Analysis Enricher

Formerly Sentiment Analysis GPT, the name has been updated to reflect that this enricher will run with any LLM selected for your environment. Now, this enricher is named LLM Sentiment Analysis. It outputs sentiment, emotion, and detects whether text is sarcastic or sincere.

Models – Create Model Button

In the Models screen, there were previously two buttons to create models. This has been changed to one button in the Inspector sidebar. When a model is selected, there is an (X) button in the top right of the inspector to close model-specific information and get back to model creation controls.

Bug Fixes

• Data Screen – Share Access Button: Sometimes, this button would not display after navigating away from the Data Screen. This has been resolved.
• Enricher – LLM Thematic Analysis: Some users experienced an error when trying to run this enricher. This has been resolved.

Improved LLM Thematic Analysis Enricher

The LLM Thematic Analysis Enricher has been improved. Now, you can add helpful context to ensure that the thematic analysis results are related to your project by giving enricher the following information:

• Research Goal
• Example Themes
• Keywords Relating to Themes

This will help ensure that the enricher’s output is relevant to your Project. The enricher will also output a justification for the theme label, as well as different options for collapsing similar themes.

Improved Persona Segmentation Enricher

Now, in the Crosstab output of the Persona Segmentation enricher, you can find an asterix (*) beside every feature used during segmentation. The features are ranked in order of importance, so your top feature will always be shown right at the top.

New Loading Icon

We’ve refreshed the loading icon!

Jobs Screen – Links to the Source and Enriched Datasets

We’ve improved the inspector bar in the Jobs screen. Now, when a job is selected you will be able to see the name of its Project. If you have permission to access that Project, you’ll be able to open the source and enriched file links displayed in the inspector screen to access them.

API – All Uploads Must be Within a Project

To ensure the best experience of the platform, we no longer support having files in the main directory—only within Projects. Now when uploading a file via API, you will need to include a Project ID.

Bug Fixes

• Stage – Copy Button: Some users reported the copy button in the list widget sometimes moving out of frame. This has been resolved.
• Enricher – Custom Theme Classification and “&”: Some users reported an issue with using the ampersand character with the Custom Theme Classification enricher. This has been fixed.

New Enricher – Persona Segmentation

This new enricher can help you find segments of populations based on a set of features in a dataset. Now, you can upload a dataset of traits, and run the Persona Segmentation enricher to understand the similarities and differences between cohorts in a given population. This enricher outputs a cluster number, along with a cross-tab file to help you understand your results.

Optimizing

Release 3.3.1 includes optimizations to ensure your environment is always running smoothly. The Q&A feature in the Stage is now faster.

New Enricher – LLM Thematic Analysis (Beta)

We’re releasing a new enricher to help you understand the themes in your data. This enricher uses the LLM selected for your environment to analyze the themes in your dataset. This is now available in the Studio screen under the Enrichment dropdown. We are interested in your feedback in this first iteration.

New Enricher – LLM Thematic Analysis – Sub-Themes (Beta)

After running the new LLM Thematic Analysis enricher, you can now break down the major themes in your dataset into sub-themes. For example, when analyzing the theme of “Billing Issues”, this enricher can help break things down further, helping you quantify sub-themes like “Invoice not sent”, “Late billing”, or “Incorrect billing”. This is now available in the Studio screen under the Enrichment dropdown. We are interested in your feedback in this first iteration.

New Enricher – Create Embeddings

To get excellent insights about your text, we use text embeddings. Previously, these could only be created by our Clustering enricher. Now, you can create embeddings using the new Create Embeddings enricher, or by using the Clustering enricher.

Data Page – Tabular View

In the Data page, we’ve added a tabular view within projects so that you can search and sort through files more easily.

Data Page – Upload data in Projects

To ensure the best experience of the platform, we no longer support having files in the main directory—only project folders.

Stage – List Widget – Filtering

Some users reported an issue with the List Widget displaying results from Ask Questions filters. This has been resolved.

Optimizing

Release 3.2.1 includes optimizations to ensure your environment is always running smoothly.

Stage – Scrollable bar charts

Previously, the horizontal and vertical bar charts only displayed the top 10 items. Now, you can choose to display more. In the stage, toggle the zoom/pan control by clicking the magnifying glass at the top of the screen. Then, use your scroll wheel on your mouse or pinch interaction on touchpad over the bar chart of interest to display more items. You can save the number of bars displayed using the save button at the top of the screen. You can also reset the widget to display 10 items by clicking the magnifying glass on top of the bar chart widget.

Stage – Normalized group-by method for bar charts

A new option has been added for group-by in bar charts. Now, you can count how many times variables in a column occur based on another variable in your dataset. You can select this “Normalized” option when creating a bar chart using group-by. This can be helpful when trying to understand questions like:

            How often does this theme occur per Submission ID?

            What is the sentiment of responses, rather than clauses?

Lab Page – Retirement

The Models page now has all functionality contained in the Lab page. The Models page displays all information relating to models, including the confusion matrixes displayed on the Lab page. This release removes the lab page from the top navigation bar.

Log-in screen – First time log-in

Some users reported issues when logging in for the first time. This has been resolved.

Stage – Reloading

Some users reported the Stage would reload when enrichers were running. This has been resolved.

Stage – List widget filtering after backspace

Sometimes, after typing text in the search bar of the list widget, search results would not refresh after removing characters from the search text with the backspace button. This has been resolved.

Log-in page – Optimized

Some users were experiencing a wait when loading the log-in page. Some also experienced a blank screen after logging in. This has been resolved.

Stage – List Widget formatting

The List Widget now displays GenAI responses with easier-to-read formatting.

Data Page – Share Access button

Previously, the Share Access button in the Data page would no longer display after a file was selected. Now, on the Data page, the Share Access button is always available in the Inspector, even when a file is selected.

GenAI Backup for Online Translator

The Online Language Translator enricher sometimes struggles to translate very small or ambiguous pieces of text. Now, when this happens, the GenAI selected for your environment will translate instead.

Jobs – Optimized

The Jobs page is now more responsive. Click on individual jobs in the table to see more details about your enrichments in the Inspector bar.

Stage – Ask Questions speed optimized

Now when using the Ask Questions function, you’ll receive responses much more quickly.

Stage – Default Dashboard size limit

Previously, when loading a dataset in the Stage for the first time, widgets would be generated for most columns in the dataset. This resulted in reduced performance on datasets with 100+ columns. Now, default dashboards will draw widgets for enriched columns, as well as the first 4 columns of the dataset

Stage – “Group By” issue resolved

Some users reported an issue when using the  “Group By” field on bar charts. This has been resolved.

Stage  – Backspace issue resolved

Some users reported an issue when pressing backspace when naming widgets in the Stage. This has been resolved.

Stage – List Widget Upgrade

In this release, the List Widget has been upgraded to make the stage faster. The filtering functionality has been moved from the tooltip to the far right of the widget. Remember, the List Widget is better at summarization tasks, while the Ask Questions function is better at searching for answers to your questions.

Stage – Unexpected reload on save

Some users reported experiencing an unexpected reload when saving reports in the Stage. This has been resolved.

Enrichment – Optimizing

We’ve been optimizing our enrichments to run at a higher speed. This helps you get to your insights faster than before.