What’s New in Discovery AI for Advantage Platform Documentation
The latest changes for Discovery AI for Advantage Platform are shown below.
|
Document |
Publication Date |
Topic |
Description |
|---|---|---|---|
| 202508.2.0 | 25 Aug 2025 | – | Bug fixes and UI improvements only |
| 202508.1.0 | 18 Aug 2025 | – | Changes to UI and error messages. |
| Searching Documents, Searching Document Text | Improved descriptions of search features. | ||
| 202507.2.1 | 03 Aug 2025 |
You can select and delete multiple extracted provisions, including fields, clauses, tables, and obligations from the Review screen. | |
|
If a Skip Review (review level 0) fails and the document is assigned a Ready for Review instead of Complete status, the Reviewed By field will reflect the name of the user who reviews the document. | |||
|
Already-processed documents can be re-extracted for additional fields or clauses without having to be re-imported from external systems. Existing agreement records can be updated or new records created. | |||
| 202507.1.1 | 21 July 2025 |
The platform administrator can use the platform's Migration Manager APIs to migrate selected objects from one org to another. | |
|
Reviewers can edit or delete one or many (but not all) pick list values from the user interface. | |||
| Page Numbers for Extracted Tables When tables are extracted, the (scanned) page number they appear on is displayed in the Tables review list item. | |||
| 202506.1.0 | 16 June 2025 | You can resubmit an already imported and reviewed document directly through Discovery AI if the first import cycle failed to extract. | |
|
Discovery AI now supports up to five image and ten text examples for field provisions. | |||
|
To help bring new users up to speed, Discovery AI now includes pre-configured NDA, MSA, and SOW mappings and worksheets with example provisions. You can directly import such documents in PDF and DOCX format with no worksheet customization required. You can also use these pre-configured custom objects as examples for building your own worksheets and mappings. | |||
|
Building custom table provisions no longer requires nominating a primary column. | |||
| 202505.2.1 | 4 June 2025 |
In addition to text examples, you can now upload images, such as a tables, signature sections, or other graphics. | |
|
Extracted fields and clauses are presented in the same order as they appear in the contract document. | |||
|
When you apply filters, you can save them as a view. You can name and describe each saved view, share them with other users, and edit or delete them. | |||
| 202505.1.0 | 19 May 2025 |
Editing a clause during review triggers clause comparison, updating similarities and differences. When the user exits without saving, the extracted clause is restored to its original state, and similarities and differences must be re-triggered to be updated. Manual clause comparison invocation can be configured on a per-worksheet basis. | |
|
Discovery AI can extract lookup fields from tables and map them automatically to corresponding CLM objects. | |||
|
Obligation extraction is now fully available. The obligation annotation type now supports multiple data types, such as lookup, pick list, etc. Reviewed obligations are saved to agreement tables. | |||
|
You can reorder, add, or delete columns in the Document Dashboard or Documents section and make it your default view. These custom views persist across sessions. | |||
|
The Discovery AI user interface has enhanced custom provision clause text provisions with support for special instructions and prompts. | |||
|
An "Other Languages" option is provided to train the AI to extract contracts in not-officially-supported languages. | |||
| 202504.1.1 | 28 April 2025 | Adding a Custom Provision | The character sets permitted in custom provision names have been expanded greatly, including Latin diacriticals and 16-bit (non-Latin) alphabets. A limited set of characters ( /, \, *, :, ?, ", <, >, and | ) remains restricted. |
| Adding a Custom Provision | The Special Instructions field now supports the pick list data type. | ||
|
The legacy document import experience from CLM is improved with persistent view settings and advanced filter criteria. Filtration now works on metadata fields for legacy document import, with filtration now working for both the Agreement Selection and Document Selection import steps. | |||
| 202503.2.0 | 14 April 2025 | Special Characters in Directory Names |
The character sets permitted in Discovery AI directory names have been expanded greatly, including Latin diacriticals and 16-bit (non-Latin) alphabets. A limited set of characters ( /, \, *, :, ?, ", <, >, and | ) remains restricted. |
| Scheduled Document List Report Generation |
Users can schedule document list report generation. This automates generating a CSV-format report of all documents and their statuses, emailed to the recipient daily, weekly, or monthly. See To schedule a list report for more. | ||
| 202503.1.0 | 07 April 2025 |
The Details page for custom provisions now offers a Special Instructions text box for users to instruct the AI to help define its scope and bound its output. | |
| The Clone feature, which can clone a worksheet's fields, clauses, tables, and insights, can now clone obligations as well. | |||
| Reviewers can now see which worksheet was used to extract information in their view. | |||
| A toggle has been added to the administrator's data extraction mapping flow to enable reviewers to extend the provisions extracted from incoming documents. | |||
| When configuring fields, clauses, tables and insights during provision mapping, users can access a bulk delete option. When setting up data extraction, users on any worksheet under any tab can bulk-delete one or more mapping row entries. | |||
| The Import Document button on the Document Dashboard has been updated to behave like the Import Document button on the Project Dashboard. | |||
| 202501.3.0 | 3 March 2025 | Prompt-Based Extraction for Custom Provisions | Prompt-based extraction offers an administrator the ability to provide natural-language input to establish custom provisions for extracting fields from a contract. This feature returns responses from your trained model and from a generative AI based on administrator prompts. |
| Phrase-Based Extractions | Discovery AI implements phrase-based extractions (Key Phrases) for custom provisions to help administrators configure custom provisions for searching and identifying field-type provisions. | ||
| 202501.2.1 | 12 February 2025 | Adding Unmapped Fields and Clauses | Discovery AI can suggest fields or clauses to map and extract from CLM for a given agreement record and offers the ability to update the worksheet. You can map unmapped field or clause extractions directly to your worksheets and use these worksheet improvements on future document imports. |
| 202501.1.0 | 27 Jan 2025 | Review History | Discovery AI now offers a Review History tab, which enables users to track all reviews and saved changes made to a document, with attached date stamps. The review history is not editable, so it is suitable for auditing. |
| 202411.2.0 | 20 Dec 2024 | Training | The Training tab, including the Force Train button, is now available during auto-extraction before the provision is trained. |
| The Home Screen | When the agreement is in Ready for Review state, the CLM web link and (if reverse sync is set up) the Salesforce agreement link are displayed. | ||
| Question Insight Options | The Insights Answer window is now resizeable, with a look and feel resembling that of the clause display. Users can now edit the AI's answers, with Markdown elements visible during edits. | ||
| To review imported documents | Reviewers can uncheck a single field or clause extraction if the extraction is incorrect. User feedback (thumbs up and thumbs down) is always enabled for single extractions, whether they are selected or not. Users can remove feedback (neutral) so that feedback won't be returned to the agreement record. | ||
| Managing Annotations | Administrators can view past user annotations/samples and remove samples as required to improve the quality of AI training. | ||
| 202411.1.0 | 6 Dec 2024 | All | Bug fix release. No new features. |
| 202410.2.0 | 22 Nov 2024 | To configure a custom provision with table extraction |
Modified documentation to match evolving UI and flows. Added Strict Name Match feature. |
| Export Annotations as CSV | A new Export button exports a provision's annotations (Document Name, Annotation Text, Feedback, User, and Date of Annotation) to a CSV file. | ||
| Direct Salesforce Agreement Links | Clicking the agreement link for a document in Complete status takes the user directly to that agreement in Salesforce. The user no longer must search for the agreement record on Salesforce. | ||
| 202410.1.0 | 8 Nov 2024 | The Project Dashboard | Revised description. New columns added to CSV/Excel exports. |
| 202409.2.0 | 25 Oct 2024 | Match Exact Column Count for Tables | Users can configure matching criteria for tables to be extracted, lessening post-extraction cleanup of unwanted tables. When the "Match exact column count" feature is enabled, only tables with a column count matching that of the configured table are displayed on document review. If "Match exact column count" is disabled, AI can match tables with more or fewer columns than are specified in the table configuration. |
| Document Review Enhancements |
Questions and answers now available for review and editing during document review. Questions can now be saved as criteria in the Agreement Insights table and as answers in the Guidance section. | ||
| Specify Table Alignment |
Users can specify in a worksheet whether the table to be extracted is aligned vertically (headings on top) or horizontally (headings on the margin). The AI can also auto-detect table alignment, though this increases processing time. | ||
|
202409.1.0 |
12 Oct 2024 |
Removed information about the summary from the "Details Tab" section and added a new "Reviewing the Summary" section under "To review imported documents". | |
|
New topic | |||
| |||
| |||
|
202408.3.0 |
27 Sep 2024 |
During document review, standard provisions and extracted provisions are compared semantically using machine learning. A new API, /semantic-comparison, is introduced. | |
|
The custom provision creation flow is enhanced with a simplified, guided flow that helps users define custom provisions correctly. | |||
|
202408.2.0 |
13 Sep 2024 |
NA |
No documentary changes. |
|
202408.1.0 |
30 Aug 2024 |
An improved user interface helps administrators select provisions when mapping fields and clauses. | |
|
202407.2.0 |
16 Aug 2024 |
A graphical indicator gives the user a quick way to assess the language model’s confidence in the quality of its extractions. | |
|
The reviewer’s table extraction experience is enhanced with the ability to select and unselect all rows in a table, to edit all rows (edit a column), and delete all rows (delete a table). Reviewers can also bulk edit/delete identical values in a column. | |||
|
The Insights feature presents a cleaner and more intuitive user interface and a Gen AI-based risk analysis option. | |||
|
Reviewers have a new ability to review extraction quality with a thumbs-up, thumbs-down widget that is monitored by the machine learning tool. | |||
|
202407.1.1 |
02 Aug 2024 |
If a field or clause is mapped to a default value, this value is now visible in the review window if it is not extracted. The value’s status as a default is also indicated. This 202406.2.0 release feature is extended to picklist-type objects. | |
|
Organization and agent details are added to the Administrator’s home screen. An Organization Role field is added to the data extraction mapping screen. These details can be used to evaluate contract risks in risk identification requests. | |||
|
Admins can set a two-digit decimal confidence threshold that the AI engine can access when processing documents. | |||
|
202406.2.0 |
12 Jul 2024 |
| |
|
Added a description of the administrative use case for the new table extraction feature. | |||
|
If a field or clause is mapped to a default value, this value is now visible in the review window if it is not extracted. If a value has a default status, this is also indicated. | |||
|
Added a tip concerning text values in picklists. | |||
|
202406.1.0 |
28 Jun 2024 |
Added description, accounting for the new Account Summary feature. | |
|
202405.2.0 |
21 Jun 2024 |
Contract Intelligence can skip older versions of documents on an import from CLM, thus processing only the latest versions of imported documents. | |
|
Contract Intelligence can now employ generative AI to extract provisioned content, enhancing its accuracy autonomously. No changes are required to users' extraction strategies, though accuracy will continuously improve. | |||
|
Contract managers can enable or disable duplicate-document imports, so that users can choose whether to accept duplicate documents during import. | |||
|
Romanian Language Support |
Contract Intelligence’s generative AI can now be trained to extract fields and clauses from Romanian documents. | ||
|
The CLM Import feature is enhanced by the addition of 24 fields that can be used to filter agreement objects. | |||
|
The existing stack of permissible custom provision types is augmented by the addition of the Table provision type. Contract Intelligence can now recognize and extract field values from ingested tables. | |||
|
Users can add template-based table rows within documents using Contract Intelligence. | |||
|
202405.1.0 |
07 Jun 2024 |
Contract Intelligence can now import documents directly from Conga CLM for the extraction of your choice. You can filter the results and apply worksheet logic to them. Updated provisions and documents are synced with CLM. | |
|
Document imports are added to the upload popup, so you can track the upload’s progress more effectively. | |||
|
Using an onscreen toggle, users can configure Contract Intelligence to skip duplicate (older) files when importing from CLM. | |||
|
Generative AI provisions mapped to a record type or worksheet are displayed under either the Missing or Found heading based on whether there was an extraction from generative AI. If found, these provisions can extract without any training (zero training). | |||
|
A generative AI toggle option is presented during custom provision creation, enabling AI for custom provision extraction. | |||
|
202404.2.0 |
24 May 2024 |
Provision descriptions have been added as popup help to the provision mapping screen’s New Field Mapping popup. | |
|
Contract Intelligence can import and extract (OCR, Auto-Extraction, or Verified Extraction) documents from CLM. | |||
|
Contract Intelligence can import documents from CLM on Conga Platform and use the Platform to reverse-sync the data to SFDC. | |||
|
202401.1.0 |
10 May 2024 |
N/A |
Improvements did not require documentation. |
|
202403.3.0 |
26 Apr 2024 |
A user (contract manager) can access a force-train button in the Training tab to train a custom provision before having reached the minimum required number of annotations. The user must submit at least one sample (a minimum of five is recommended) and can force-train on untrained or trained provisions. Force training relies on all samples submitted by the user, not only on new samples since it was last trained. | |
|
If this option is enabled, a reviewer can choose to get the next available batch of documents for review. A pop-up window displays the list of record types that the reviewer has access to. Requests are processed in the order requested. Users cannot request a new batch until currently assigned documents are reviewed or reassigned. | |||
|
When importing a document, in addition to selecting the document type, a user (contract manager) can now select a worksheet to apply to the import, improving the relevance of the extraction. The default worksheet is preselected but can be changed through a drop-down menu. | |||
|
A user (contract manager) can detect and skip duplicate documents from a single or bulk import. During import, a duplicate file check is made within each folder and across all folders and sub-folders. An upload popup dialog shows when a duplicate file is discovered, by name. | |||
|
202403.1.0 |
20 Mar 2024 |
Updated edit and delete flow information. | |
|
Bulk resubmit now permitted on bulk uploads that are hung in processing. | |||
|
202402.2.0 |
06 Mar 2024 |
Importing Documents and Files |
Folder upload flow improved with an Uploading Documents pop-up, displaying uploaded files, files remaining to upload, upload progress, and success and error statuses. |
|
February '24 |
07 Feb 2024 |
The procedure for adding publisher APIs has been simplified to a one-time, single-token-per-tenant operation. See To set up the publisher API for details. | |
|
January '24 |
10 Jan 2024 |
Documentation added for enhanced Advanced Search feature. | |
|
December '23 |
06 Dec 2023 |
Contract Intelligence supports deploying and training custom provisions. | |
|
October '23 |
04 Oct 2023 |
Users can now bulk-upload folders from the UI without having to stage them in ZIP files. | |
|
June '23 |
08 Jun 2023 |
Contract Intelligence displays a comparison between the extracted/identified clause and the language of the standard provision. | |
|
Users can choose to extract only matching clauses or the whole paragraph in which the clause is matched. | |||
|
Administrators can configure data extraction mappings to skip a review. | |||
|
February '23 |
08 Feb 2023 |
New feature | |
|
New feature | |||
|
Product name change |
Product was named "Contract Intelligence for CLM." Product is now named "Contract Intelligence." | ||
|
September '22 |
07 Sep 2022 |
New feature | |
|
New feature | |||
|
New feature | |||
|
Added detailed compendium of provisions. | |||
|
Imported IDD REST API documentation from Max project. Added APIs for bulk upload, insights, and training. | |||
|
May '22 |
08 Apr 2022 |
All topics |
New product |
