Most companies have no idea what’s in their contracts. It’s a shocking thought when you consider how important contracts are to the many ways businesses operate, from vendor relationships to client agreements and compliance obligations.
The problem is that once each contract is signed, it goes into a repository. Eventually, in that repository, you end up with thousands of key data points around risks, obligations, and opportunities—all locked within endless PDFs. This is an unstructured data problem that can cause revenue leakage and chronic mismanagement of risk.
We want to help change this narrative. Our top priority for Relativity Contracts of late is to explore how we can leverage generative AI and output validation to help customers efficiently and accurately transform thousands of signed agreements into tables with fielded data—surfacing those insights for simpler, more straightforward interrogation and understanding in a single place.
In late 2024, we released our essential fields model which uses generative AI to auto-populate more than a dozen key fields like agreement type, parties, effective dates, and more. With this model, customers can extract key insights across many foundational fields in a lightning-fast way with a single click.
In Q2 of 2025, we launched custom prompt models in Advance Access, allowing select customers to harness the power of LLMs to extract key data points beyond the scope of essential fields. Teams at KPMG US, Cimplifi, and PwC UK have been testing out this new capability and sharing their feedback with our product team.
Recently, we had a chance to chat with these customers about their insights and expertise on contract review in RelativityOne. This article is the fruit of that conversation, featuring insights from:
- Sean Hora, Director – Forensic Technology Practice at KPMG US
- Charles Post, Executive Vice President at Cimplifi
- Rachael Hunter, Senior Manager – Legal Technology at PwC UK
- Rishi Khullar, Group Product Manager at Relativity
Data Challenges in Contract Review Projects
What are some of the top challenges you face in contract review projects?
Rachael Hunter: One of the core challenges is how to manage large volumes of complex documents. Many of our projects involve thousands of contracts across multiple entities, jurisdictions, and formats, each containing nuanced legal language that requires careful analysis.
Another major hurdle is the lack of contract visibility. Clients often don’t have a structured repository of their contracts, making it difficult to quickly assess obligations, risks, and renewal timelines. This fragmentation complicates both strategic planning and operational execution.
We also operate in an increasingly stringent regulatory environment. Whether it's data privacy (e.g. GDPR, DPA), ESG disclosures, or financial compliance, the bar for legal and contractual transparency continues to rise. Keeping pace with these evolving standards, while adapting review processes accordingly is a continual challenge.
Lastly, all of this contributes to a heightened risk of non-compliance. When reviews are conducted manually under tight timelines, or without the right tools in place, there's a real danger of missing key clauses, failing to spot deviations, or overlooking red flags that could expose the business to legal or financial penalties.
What are the main advantages of using Relativity Contracts to address some of these challenges?
Sean Hora: Relativity Contracts offers several advantages that directly address these challenges. Its AI-powered clause and data extraction capabilities allow teams to quickly locate critical terms such as termination clauses, payment terms, and liabilities. The platform supports both pre-built and custom AI models that understand context and structure data efficiently, significantly reducing manual review time. Additionally, its contract viewer interface and workflows enhance project management, collaboration, and reporting.
Charles Post: There are very few post-signature contract analytics platforms capable of handling contract volumes at enterprise scale. Relativity Contracts stands out as one of the most powerful platforms on the market—able to process over a million contracts using LLMs. It's a true benchmark in the industry.
How do you tailor your services to the unique needs of your clients and their specific feedback?
Sean Hora: Our services are tailored through a flexible and responsive approach that incorporates client feedback into workflows. For example, prompt criteria in generative AI models are refined based on client input to improve output quality. Our tailored approach has supported a range of use cases, such as harmonizing global contract templates and analyzing pricing language in thousands of engagement letters.
Charles Post: With over a decade of experience, we’ve developed deep subject matter expertise across procurement, financial services, M&A, life sciences, entertainment, and more. This allows us to not only anticipate client needs, but to rapidly align our analytics frameworks with specific industry use cases and feedback loops. Client input is essential to our process, but our proven frameworks mean we never start from scratch.
Rachael Hunter: We start with a client discovery phase, understanding their key challenges, document portfolio, key data points required, and impacting business priorities. This allows us to design a review taxonomy that’s fit for purpose, often involving custom data rules, generative AI prompts, data enrichment steps, and logic trees for issue tagging and escalation. We typically test this approach on a sample or training set to validate and refine the rules before deploying at scale, ensuring the framework is robust, targeted, and aligned to client expectations.
In our projects, we build flexible frameworks that can be refined mid-review based on client feedback. That feedback loop, combined with robust QA and validation stages, ensures alignment with client priorities and desired outcomes.
Prior to generative AI, what was your workflow for executing contract review projects?
Rachael Hunter: Before the introduction of generative AI, our contract review workflows relied heavily on rule-based approaches and manual effort. We used a combination of regular expression rules, keyword searches, and machine learning models to identify relevant clauses and terms across the large document sets. While these tools provided a reasonable level of automation, they often required extensive tuning, bespoke data normalisation scripts, and didn’t always capture the nuance or variability of legal language.
At scale, for most reviews, the delivery depended on large review teams. Human reviewers played a critical role in interpreting context, applying judgment, and handling clause exceptions that automated tools couldn’t reliably address. This model offered control and quality, but it was resource-intensive and challenging to scale efficiently under tight timelines or changing client needs.
Current Generative AI Capabilities in Relativity Contracts
What benefits have you seen from Contracts’ essential fields model powered by generative AI?
Sean Hora: With the essential fields model we have been able to provide clients with information on their contract population very quickly, which has driven confidence in the tool’s capabilities. It reduces the need for manual tagging and enables faster project turnarounds, particularly in high-volume reviews. The ability to show the speed and accuracy of the out-of-the-box generative AI capabilities with the essential fields model has allowed our team to instill client confidence in the custom prompt capabilities as well.
Charles Post: Accurately extracting core bibliographic data like effective dates, counterparties, and expiration terms is deceptively complex. Relativity Contracts handles it with precision—turning what was once a manual QA headache into a reliable, automated extraction.
What benefits have you seen from Contracts’ custom prompt models capability that is currently in Advanced Access?
Rachael Hunter: One of the biggest challenges in contract review is dealing with the variety and complexity of legal documents, with differences in structure, language, and formatting that can make standard automation approaches less effective. Approaches such as custom prompt models enable us to build and deploy tailored extraction rules that perform well across diverse document sets. This flexibility helps address variations in language or structure, supporting accuracy and speed even with complex contracts.
Why are custom prompt models important to your strategy for contract review services?
Sean Hora: Custom prompting within RelativityOne is central to our contract review strategy because it combines the capabilities of Relativity with those of generative AI in a highly effective way. Every contract review project we do involves the identification and/or extraction of custom contractual elements. The ability to customize prompts gives us the ability to be even more agile and responsive to client needs.
Charles Post: Custom prompting plays a key role in our shift from billable hour models to fixed-fee, per-document engagements. It allows our subject matter experts to directly shape analysis using low-code tools—capturing their expertise into repeatable, scalable processes.
How have clients responded to the use of generative AI for contract review projects?
Sean Hora: Clients have responded positively to the integration of generative AI in contract review and have also shown interest in expanding use cases, including pricing analysis and template harmonization. Our clients appreciate the speed, flexibility, and depth of insights provided by AI-enhanced reviews.
Charles Post: Adoption often begins gradually—because meaningful change takes time. But early adopters quickly see a competitive advantage: faster response times, deeper insight, and scalable ROI. Once they see results, the conversation shifts from "why AI?" to "what else can we do with it?"
Rachael Hunter: Clients have responded positively to the use of generative AI in Contract Reviews, particularly in terms of faster turnaround times and the ability to identify key terms and risks across highly varied data sets. To ensure reliability, we adopt a “human-in-the-loop” approach, conducting sample checks for quality assurance and manually reviewing edge cases. This hybrid model balances the efficiency of AI with the expertise of human reviewers, giving clients confidence in both the technology’s capabilities and the robustness of the final deliverables.
Improvements Coming to Generative AI Capabilities in Relativity Contracts
What enhancements are you making to the essential fields model and custom prompt models?
Rishi Khullar: We intend to improve these capabilities across several dimensions: speed, simplicity, accuracy, and document size. We plan to simplify and modernize our architecture and the end-to-end user experience, as well as include more prebuilt fields that will come out of the box with these capabilities. We are really focused on increasing the speed to value—piles of PDFs to tables of fielded data—and our roadmap is dialed in on this metric.
As we upgrade to newer versions of the foundational models powering these capabilities, we’ll continue to support larger documents and improve the accuracy of outputs. We perpetually iterate and experiment with our underlying prompts on the latest models to get the most accurate possible results.
What other problems will you solve with generative AI in the future?
Rishi Khullar: There are too many opportunities here to list! We’re seeing promising results as we constantly explore how AI can solve problems in contract review.
One such problem is around contract families. Customers will often receive a flat tranche of documents without any grouping of master agreements with their amendments, addendums, exhibits, or statements of work. This is a painful manual review effort; then, even once you have the family groupings, there remains the problem of getting insight into the latest and greatest controlling, active terms across the whole family, not just within each document. We’re experimenting with AI approaches to attacking this problem, one which the market is not currently solving well.
There are plenty more painful problems in contract review, and our experimentation is driving optimism that generative AI will help us solve them.
Take a Closer Look at Relativity Contracts
You can explore Contracts for yourself at Relativity Fest Chicago this October. Be sure to add session, "Relativity Contracts in Action: Real-World Use Cases & Lessons from the Field," to your agenda, and ask someone from the Relativity team to give you a closer look at the tool. We'll be happy to talk all about it!
Graphics for this article were created by Sarah Vachlon.
