Accelerate conflicts analysis: 3 phases for getting started

Does your firm’s conflicts process take too long and yield unclear results? Analyzing and escalating conflicts issues are the most time-consuming parts of a conflicts clearance process.

The AI-assisted conflicts clearance in Intapp Conflicts accelerates this step, providing conflicts analysts with results organized by the level of scrutiny required —enabling analysts to quickly complete their reviews.

Our AI-assisted conflicts clearance function uses machine learning AI. Intapp AI-assisted conflicts focuses on analyzing patterns and making predictions based on your firm’s structured data, previous predictions, and analysts’ feedback. This process refines the issue classification over time. The inputs are the firm’s search results, and the outputs are suggestions about which hits are more likely to be issues versus non-issues using feedback from your own firm’s users.

The model improves over time and retrains weekly to stay aligned with your business as it evolves. This continual improvement further accelerates conflicts review and clearance over time.

Moreover, with Intapp Conflicts, your firm’s data stays in place because the AI model powering Intapp Conflicts is part of your firm’s Intapp tenant infrastructure.

Firms can begin using Intapp Conflicts with minimal disruption to their current conflicts process. Below are several considerations for configuring Intapp’s AI-assisted conflicts clearance function.

1. Process considerations

If your firm is moving to the cloud, where you can take advantage of AI-assisted conflicts clearance, there are several process issues you can start thinking about.

• What specific pain points do you aim to address? For example, are you trying to speed up the conflict-checking process, improve accuracy, or reduce human error? Identify the use cases where AI assistance can bring the most value. The AI model, which is the repository where summarized patterns of past searches are stored, can then be trained with data points collected from past hits in those searches and firm-specific rules for classification of risks implicit to the data points available in the Intapp Conflicts database (such as analyst decisions) to make more accurate predictions — that is, recommendations — about new hits, streamlining conflicts evaluation. For instance, if your firm’s strategic objective is to limit taking on work for current or potential clients within a specific industry, a “high-risk rank” conflict rule can be designed to prioritize such results in the conflict search in the “issue” category, thereby alerting analysts.
• Does your conflicts review process include closing searches on a consistent basis? This is relevant because the AI is trained once a search is closed — without this step, the AI model can’t learn from the previous search, and as a result, it won’t be able to refine its recommendations based on the feedback or outcomes of the search. Closing searches consistently ensures that the AI has the necessary data to improve its pattern recognition and make more accurate predictions for future conflicts searches. If searches aren’t properly closed, the AI lacks the crucial feedback loop that helps it continually evolve and better align with your firm’s specific needs and judgment, ultimately limiting the efficiency and accuracy of the conflict clearance process.
• Do your conflicts review and clearance processes include consistently applying resolutions or selecting report hits? One of these options can be selected as a configuration option for baseline training. Consistently applying resolutions or selecting report hits during the conflicts review and clearance process is a crucial step for baseline training of the AI. This process helps the AI to learn and adapt to your firm’s specific practices by providing a consistent feedback loop on which hits are relevant and which are not. By regularly applying resolutions (for example, via applying appropriate resolution selection to a hit, such as the “No conflict” or “Waiver acquired” resolution) or selecting report hits (via “Add to Report Hits” tagging), you ensure that the AI model receives supplemental data to refine its decision-making and improve the accuracy of its future recommendations. This consistency is essential for maintaining the quality and effectiveness of the AI-assisted conflicts clearance process over time.
• Are your existing conflicts rules helpful? Intapp Conflicts enables firms to establish their own specific conflict rules, which can be used to filter or prioritize results based on predefined search parameters. These conflict rules are incorporated into the AI model and contribute to its training, offering immediate advantages in the conflict-checking process.
  
  You can also map certain rules to an AI category for foundational training. For instance, if you have a rule to rank a hit where a prospective client’s name matches a counterparty involved in an active matter, you can map that type of hit to always appear in the “issue” category.
  
  Evaluating and enhancing your existing conflicts rules before you start using AI-Assisted conflicts will help improve recommendations.

2. AI training considerations

• How much data do you have? At least 300 closed searches containing a minimum of 250,000 search results across them are optimal for initial training of the AI model. This number varies based on the input of the initial training and the consistency of the conflicts team who close searches. A model whose initial training is based on a small dataset (that is, very few searches) with hits resolved or exported to a report will result in a less robust model than a model with a larger training dataset.
• Who will train the AI model? It’s important to define roles and associated system capabilities clearly within your conflict-checking process. The AI model’s initial training can be configured by an implementation resource at the time of the AI-assisted conflicts deployment. Firms with conflicts team members who have different skill levels can designate specific members of the team to be responsible for the initial AI model training. Members of the team who are involved in AI model training can use suggested AI Conflicts view for analyzing results, while the remaining team members can continue using the classic interface in the process. Keep in mind that while Intapp AI may perform the initial scan and identify potential issues and what hits could be excluded, human oversight is still necessary to evaluate the context and nuances of specific situations. 
• Do you prefer to use hit resolutions or hits included in the prepared conflicts reports for initial training of the AI? The AI model initial training supports one of two options: “Train on past hit resolutions” or “Train on hits that have been included in conflict reports.”    
• Which resolution types should the system treat as issues or non-issues for ongoing learning? During AI Model setup, the firm’s Resolution types are mapped as either “Issue” or “No Issue.” The AI engine then uses these classifications to inform its recommendations on future hits, contributing to its ongoing learning process. 
• Are there any search types that should be excluded from ongoing AI model learning? When configuring the AI model, you can exclude certain types of checks from AI model learning. For example, preliminary conflict check search types can be excluded from the AI model learning.  
• Which conflicts rules should be treated as issues regardless of the system suggestion? You can configure the AI model to map hits for certain conflict rules to categories. For example, you may want to ensure all black books appear in the Issue category for closer scrutiny. 

3. Configuration considerations

• What kind of professional services do you want to work with? You may want to select a provider who will configure the training model, user interface, and role/user capabilities. You could also have your service provider set an initial training approach based on the AI training considerations listed above. Once configuration is complete, your professionals can use Intapp’s AI-assisted conflicts clearance function while the model continually trains in the background.

It’s also important to continually monitor the effectiveness of the AI’s output. Regularly assess the accuracy of the model’s recommendations as well as their impact on the firm’s efficiency.

• Are real conflicts being identified faster?  
• Are the results of the AI tool aligning with human judgment?  
• Are there areas where the system could be improved?  

Feedback from the conflict checkers and those involved in conflict clearance is valuable for refining both the technology and the process. Continual improvement ensures that the AI-assisted conflict-checking system remains aligned with your firm’s evolving needs. Thoughtful planning and process integration will help create a conflicts system that enhances efficiency, reduces errors, and improves overall outcomes for your conflict-checking efforts.

Implement AI-assisted conflicts review

AI-assisted conflicts searching offers numerous benefits to firms:

• It reduces the volume of hits analysts need to review
• It displays results in a way that makes them easy to analyze
• It removes irrelevant erroneous and “false-positive” results
• It provides reasoning for its classifications

The AI recommendations engine takes an initial pass at categorizing potential conflicts hits, labeling them as “issue,” “no issue,” or “uncertain.” Your conflicts analysts can then prioritize their time reviewing hits that require resolution. At no point will the AI model make any final decisions about a hit or whether a real conflict exists.

Over time, as your analysts engage with Intapp Conflicts, the model learns from users’ acceptance or rejection of the suggestions. As a result, the AI improves, with its output of potential conflicts consisting more and more of hits that are actually issues the firm should review. This improvement in the AI accelerates conflicts review and clearance moving forward.

Ready to take the next step in optimizing-checking process? Reach out to Rabiya Hirji (Sr. Solutions Advisory Director) or Yelena Chervinsky (Director of Risk Consulting) for further guidance on getting started with AI-assisted conflict checking or to explore our consulting services. We’re here to help ensure your firm’s success in leveraging Intapp’s powerful AI-assisted conflict-checking capabilities.