Provision models in the Provision Library are classified as built-in or custom. While built-in provision models cover a broad range of use cases, they may not be suitable for every document. In such cases, you can define your own custom provision model, add examples, and train Discovery AI to extract it.
Figure 1. Custom and Built-In Provision Models
Table 1. Provision Model Characteristics
Category
Built-In Provisions
Custom Provisions
Definition
Pre-defined and pre-trained by Conga, available out-of-the-box in Discovery AI for commonly used contract types.
User-created provisions designed to capture organization-specific or uncommon contract data.
Coverage
Over 1,350 standard legal clauses and data points across 130+ document types.
Customizable to any use case.
Training
Ready to use without additional setup or model training.
Requires configuration, including at least two sample documents, to train the model.
Model Source
Pre-trained by Conga's AI and legal experts.
Admins can refine model performance by reviewing and incorporating reviewer feedback, entering special instructions to refine results, and offering prompts or key phrases for more accurate and comprehensive extraction.
Actions
Cannot be edited or removed.
Can be modified or deleted by administrators as needed.
Built-in and custom provision models are further categorized into fields, clauses, tables, and obligations.
Fields are measurable data such as dates, duration, percentage, amounts, pick list items, or short text.
Clauses consist of legal language, such as stipulations, declarations, and legal clauses such as termination or jurisdiction clauses.
Tables contain structured multi-row data like line items, prices, product lists, or delivery schedules.
Obligations are deliverables, responsibilities, or commitments that must be fulfilled under a contract.
Once a document is imported and processed, Discovery AI extracts relevant information and presents it to the reviewer in these categories.
To sync the Provision Library
Click the Provision Library tab in the left navigation bar to open the Provision Library.
Click the Sync Library button.
Discovery AI syncs all available provisions. This can take a considerable amount of time.
Built-In Provisions: Standard Provisions
Custom Provision Models
Custom provision models offer a flexible way to extract and structure data from contracts and related documents. When a contract's provisions do not align with any of the predefined, built-in options, you can define your own by naming, configuring, populating, and training the AI on a custom provision type.
Each provision must be assigned a unique name and an associated data type. Supported data types include Date, Organization, Currency, Duration, Number, Percent, Picklist, Short Text, Multi-Picklist, Text, Table, and Obligation. These data types represent CLM fields that Discovery AI uses to classify and map extracted content from contracts and supporting documents.Figure 2. Custom Provision Model Structure
Creating Custom Provision Models
To create a custom provision model in the Discovery AI Provision Library:
Access the Admin Console. From the Conga Start window, go to Admin Console > Discovery AI > Admin Dashboard.
The Discovery AI Admin interface opens to its home page.
Select the Provision Library tab from the left navigation panel to access provision models.
Carefully review the list of available provisions to ensure that the one you need does not already exist.
If the required provision is not available, click NEW PROVISION to open the new provision creation pop-up.
Enter a unique name for the new provision.
Latin and non-Latin character sets are supported.
Select the annotation data type from the available options: Date, Organization, Currency, Duration, Number, Percent, Picklist, Short Text, Multi Picklist, Text, Table, or Obligation.
Note:
The Picklist, Multi Picklist, Table, and Obligation data types offer additional configuration settings.
Tip:
Best Practices for Creating New Provisions
Name: Provide a clear, natural-language name for the provision. Use the standard legal term for accuracy and avoid abbreviations, numbers, or symbols. For example, use "Confidentiality Clause" rather than "Confidentiality_Clause_SLA_Project" or "Confidentiality_v1" etc.
Annotation Data Type: This field defines the specific type of information (also known as an annotation) that the system is designed to extract from a provision. The data type determines the kind of value the field can contain, such as text, number, date, or picklist, or more complex structures like tables and obligations. Select the appropriate data type here for accurate extraction. For clause extraction, select the Text data type.
Description: This field outlines the meaning and intent of the field, clause, table, or obligation to be extracted. It directs the AI model toward the correct information. Provide a clear, specific description rather than the actual clause text.
Maximum Extractions to Display for Review: The value in this field indicates the number of extracted instances of the field or clause to present to the reviewer.
Click Next.
In the Provision Example window, click ADD EXAMPLE.
If you chose a field-type value, the Add Example pop-up offers a choice between text and image examples. If you did not select a field (i.e., a clause-type value), enter or paste a paragraph of text and the data you want to extract. For field (not clause) values:
Enter or paste a paragraph of text and the data you want to extract; or
Upload a visual example (e.g., the value within a signature box, exhibit, or table). You can drag and drop an image or upload it from your desktop.
Provide instructions to extract a field value from a specific table or page.
Click Add another example and repeat the previous step.
Note:
At least two examples are required to train the AI. For fields, you can provide up to five images and ten text examples. For clauses, you can provide up to five text examples.
in the Special Instructions field, you can enter any instructions you think will help the AI engine find the provision you are looking for (search and recognition cues) and refine the extracted results (for example, to convert text to numbers or to change the date format for localization) . See Adding a Custom Table Extraction Provision Model for more.
You may select the optional INPUT PROMPT tab to enter a natural-language description of the data you want to extract.
This command prompt enhances extraction accuracy and operates independently of the standard AI pipeline.
For example:
"Extract the agreement's total contract value."
"Retrieve the contract's effective duration in calendar days."
You may select the optional ADD KEY PHRASES tab to enter specific keywords or phrases that refine the AI's extraction process. Use distinctive terms rather than generic terms like "party" or "order". The AI prioritizes extractions containing these keywords through exact and fuzzy matching. Add key phrases only if the examples and instructions alone do not produce the desired extraction results.
Click DONE.
A success message appears and a summary of the new provision is displayed, confirming its creation.
Custom Provision Model Creation Examples
Agreement Start Date
Create a custom provision model for an Agreement Start Date field
Train the system by adding at least two examples. In each example, include a passage containing the field you will extract in the Paragraph Text field and the desired field data alone in the Data To Extract field. Figure 3. Training the System You can also provide an image with data extraction and instruction examples for training. Use special instructions to fine-tune or format the extracted information, or to give the system additional extraction instructions if the examples do not extract the desired information.
Create a Custom Severability Clause Provision Model
To train the system for clause extraction, add the clause text and required data to extract in text format.
Order Form Table Provision Models
Custom provision models with table annotation data types are ideal for contracts with tabular data, such as order forms or product or service listings. Table extraction uses generative AI that requires no training. After you configure it and map the worksheet, data can be extracted from properly structured tables with clear headers and aligned rows.
Adding a Custom Table Extraction Provision Model
You can use AI to extract tables from documents. This is especially useful when contracts present such tabular data as bills of materials or delivery schedules. Because table extraction requires generative AI, these extractions do not require training. Users can extract tables as soon as the table extraction is configured and mapped to a worksheet.
Each table must have a columnar structure, with each line reflecting the values named in the heading. Discovery AI extracts these values in much the same way it handles fields and clauses, presenting this data in a column-defined line-item format.
Select Table from the Annotation Data Type drop-down menu.
Enter a name for the custom provision in the Name field, select a language, and enter a brief description in the Description field.
Select a table by name from the Choose Table drop-down menu.
These tables are defined in CLM. You can set up new tables in CLM by following the instructions for adding contract line items.
Click NEXT.
Click ADD COLUMNS to raise the Add Alternate Keywords for Columns pop-up.
Select a column name from the Choose Column drop-down that you want the AI to extract.
Click the NEW COLUMN button to add columns to extract. Add columns in the order you want this information presented. The order you see here is the presentation order of the extraction, irrespective of the order the columns fall in the scanned table. You are not required to select every column.
Note:
The order of the columns here controls the order in which they are presented to the reviewer. To change the order, you must delete and add columns until you achieve the desired order. It is therefore a good idea to have the presentation order in mind before you begin.
For each column, enter keywords as comma-separated text in the Add Values for Keywords field approximating likely column headings to enhance the AI's likelihood of accurate extraction. For example, if you know a column normally titled "Description" is occasionally titled "Desc.", you can enter that alias here.
You can check the Strict Name Match box to enforce verbatim column-heading matches. Leaving this unchecked permits "fuzzy" (close) AI matching.
If the table type contains hierarchic data, indicated by regularly indented rows in existing columns, you can set a hierarchy level for a given column.
For example, if a table has a column named "Prices" with sub-entries for "$1–$5,000", "$5,001–$10,000" etc., you can pick the Prices column from the Choose Column and assign it to Hierarchy level 1 in the Type field, then repeat the process for the price categories, identifying them as Hierarchy level 2. Selecting hierarchic values deactivates strict name matching and keyword values, and disables the Add Values column. A hierarchy type only specifies hierarchy level and therefore cannot be combined with specifying keywords.
Note:
Do not enter values with commas except to divide entries. Entries in the Add Values for Keywords field are comma-delimited, so your entries will be broken into fragments at each comma.
You can slide the Match exact column count toggle to ignore tables with an unexpected number of columns.
You can configure the table definition for "hybrid" tables in which horizontally or vertically ordered information is presented in a column. Your choices are Auto Detect, Column Data, and Row Data.
Auto Detect is the default selection, and is usually the best. If you expect a table column to contain cells with multiple sub-columns, select Row Data. If you expect a table column to have cells containing multiple vertically-ordered rows, select Column Data.
CAUTION:
If you select either Column Data or Row Data for one column, you must apply a Data Orientation value to all rows or columns in the table model.
For more on hybrid table model configuration, see Hybrid Tables.
Click the SPECIAL INSTRUCTIONS tab to enter additional instructions to the AI engine. Use natural language, avoiding jargon and excessive abbreviation, to add any instructions you think will help the AI find the provision you are looking for (search and recognition cues), and to describe how to present extracted table data (formatting preferences).
Tip:
Be specific about how to extract table data.
If a formatting preference applies to a specific column, use that column's name as a reference.
"Transpose the column that lists year-wise data into row-wise data corresponding to the Program. List each year in a column named "Year". Flatten the merged cells (replicate eight times for eight years) in the Product column."
"Flatten the merged cells in the first and third columns."
"The headers in columns 2 and 3 are nested. Flatten "Base Rebate Percentage" to its two nested sub-headers."
When you have selected and described all columns desired for extraction and added any special instructions to help the AI learn what you want, click ADD to review your columns.
Click DONE to save the new custom provision model and return to the Provision Library.
