As eDiscovery datasets grow, finding the right information can take time. Even when documents are organized and searchable, identifying connections, understanding conversations, and locating potentially important evidence can require multiple searches and careful review.
Relativity aiR Assist offers another way to explore the documents in a workspace. Using natural-language questions, users can ask about the information contained in indexed documents and receive answers supported by citations to the underlying evidence.
But how does it work, what can it help with, and what should users know before getting started? Avalon’s Director of Legal Tech Operations, Josh Burke, breaks down the essentials.
“One of the biggest advantages of aiR Assist is that you don’t have to know exactly which keywords to search for to start exploring your data. You can ask a question in plain language and let Assist help identify documents that contain the information you’re looking for.”
aiR Assist lets you ask questions about your indexed documents in everyday language instead of building traditional keyword searches. Simply enter a question, and Assist identifies relevant documents and provides an answer with citations so you can review the original information and see the context behind the response.
“Assist isn’t just giving you an answer and asking you to take it at face value,” said Burke. “Its citations let you go directly to the source documents and evaluate the information for yourself.”
For example, instead of running multiple searches to investigate a particular conversation, a user might ask:
“What did Alex discuss with Jordan about the contract between January 1 and March 31?”
Assist can use the question to identify relevant documents and synthesize the information into a citation-supported answer.
The goal is to make it easier to explore a document collection, identify potentially important evidence, and develop a better understanding of the information within a matter.
One important consideration is that Assist can only retrieve information from documents included in its index. Each index can contain up to 300,000 documents, with a maximum of five built indexes per workspace. Where applicable, Avalon maintains a primary index called “All Documents,” which is rebuilt after each new document load.
However, not every document in a workspace is automatically eligible for indexing. For example, documents must:
“It’s important to understand that Assist can only work with what’s been indexed,” Burke explained. “If a document doesn’t have extracted text, falls outside the saved search, or isn’t available to the user, it won’t be part of the results. Knowing what’s included in the index helps set the right expectations. But, if additional visibility is needed, Avalon can provide a saved search identifying documents excluded from the default index.”
Additional indexes can also be created for specific requests or custom workflows. When multiple indexes are available, users simply select the one they want to query. Understanding these parameters is important because the quality and scope of Assist's results depend on what information is available in the index.
Yes, but there’s an additional step. By default, Assist searches document text. When metadata mapping is enabled for the workspace, users can turn on the “Use Metadata” option when submitting a question. This allows Assist to apply mapped metadata filters before conducting its semantic search.
Current available fields include (Relativity is working to add more in the future):
For example, if a user wants to understand communications between two individuals during a specific period, metadata can help narrow the search to the relevant participants and dates before Assist retrieves matching content. This can be especially useful when investigating communication patterns or trying to understand who was involved in particular discussions.
The possibilities extend beyond finding documents that mention a particular keyword. Assist can help users explore themes, identify relationships, understand events, and locate potentially important communications.
“I encourage users to think beyond traditional keyword searches,” said Burke. “You can ask Assist to help identify relationships between people, understand the sequence of events, or find communications around a particular topic. It’s another way to explore the documents and uncover information that might otherwise take several searches to find.”
Some common eDiscovery use cases include:
Finding potentially important documents:
“Can you find me documents that discuss [topic]?”
Identifying themes or potential misconduct:
“Are there any documents mentioning fraudulent behavior by [individual]?”
Understanding who was involved:
“Who participated in discussions about offering [topic]?”
Building a high-level timeline:
“Create a timeline of events that took place before the start of [project].”
Exploring communication patterns:
“Who communicated with whom about [topic]?”
Identifying potential exhibits:
“List documents that may be useful as exhibits based on [key criteria].”
Confirming whether a conversation took place:
“Did [individual] send an email about [topic]?”
These types of questions can help legal teams investigate specific issues, explore information they may not have considered, and identify documents for further review.
Like any technology, the value of Assist depends in part on how it’s used. A few simple practices can help users get more meaningful results.
“I look at Assist as another tool in the eDiscovery toolkit,” said Burke. “It can help you get to relevant information faster and explore your data in different ways, but it doesn’t replace traditional search, attorney judgment, or a thorough review process. Knowing when and how to use it is what makes it valuable.”
Relativity aiR Assist gives legal teams another way to interact with their eDiscovery data. By making it possible to ask questions in everyday language and receive answers grounded in indexed documents, it can help users investigate issues, identify potentially important evidence, and better understand the information in a matter.
The key is knowing how to use it effectively and understanding where traditional search and review workflows remain essential. With guidance from Avalon’s eDiscovery professionals, including Burke, teams can better understand how tools like Assist fit into their workflows and how to use them effectively.
Want to learn more about Relativity aiR and what it can do for your eDiscovery workflow? Contact Avalon to talk with an expert.