The Agiloft AI Core allows you to extract key terms and clauses from contracts. You can use a machine learning action to map these values to fields in records, as well as to create records for each identified clause in a table of your choice. You can also use this feature in conjunction with the Word add-in, but must have the Agiloft KB properly configured first. The exact setup for these use cases depends on your specific needs and your current KB configuration. You can use the information here as a starting point or introduction to the available methods of integrating AI with the rest of a KB, but be prepared to make adjustments to your own system.

AI features require the Advanced or Premium edition of with the AI Platform add-on. After you finish Setting Up AI, you need to integrate it with other tables to make it useful. It's likely you'll need to have several tables and fields already available in your system to correctly map the AI inputs and output fields.

Extracting Content

AI capabilities determine the models you can use with your system. Agiloft has an emphasis on models that are used for extracting content from documents and files, but your choice of model depends on the content you want to extract. For example, some extract metadata, such as titles and dates, while other models extract whole clauses. Some models don't extract data, and instead compare similarities or make classifications.

When it comes to extracting content, there are some limitations. For a list of other limitations, visit Limitations of AI.

Contracts AI

Contracts AI is an AI Capability, developed by , that consists of two ATHENA models that can extract information about various different types of contracts. 

When you create a machine learning action that uses a model that extracts data from the contract document, such as either ATHENA model, you can pick and choose which data actually gets extracted and which data you want the model to ignore. You can make these decisions from the Labels tab of the machine learning action wizard, as long as the model you've selected contains a label set (see step 8 in Extracting Content checklist in the previous section). If your model does not contain at least one label set, the Labels tab does not appear. If either ATHENA model is selected, the Labels tab is visible contains labels for generic data that is commonly used throughout many kinds of contract types, as well as specialized labels for data used in specific contract types. In the example below, you can see that ATHENA-CE-AS has generic labels for clause data such as Assignment and Contract Title, as well as labels for specialized or contact-specific clauses, such as Closing Adjustments for an Asset Purchase Agreement.

Although some ATHENA labels may only be relevant when analyzing specific contract types, the generic labels can be used with virtually any kind of contract as long as the document format does not drastically stray from the standard. Currently, there are many different contract types and labels directly supported by the ATHENA models.

Agiloft Contract Assistant for Word

You can also use AI to extract clauses and key terms from Microsoft Word documents directly using the Agiloft Add-in for Microsoft Word. The add-in uses preconfigured Machine Learning actions to extract clauses and metadata from the Word document, which can then be edited and added to your KB as records. You can select which action buttons to use for the Word add-in in the Word Add-in Integration wizard. For more information about the Word add-in, visit Agiloft Contract Assistant for Word.