From internal chats to collaborative documents, communication has always left behind a trail of discoverable data. But now, with the rise of generative AI, there’s a new kind of conversation emerging: between humans and machines.
Chatbots like OpenAI’s ChatGPT and Google’s Gemini are no longer novelty tools. They’re embedded in daily workflows across industries. Product teams are drafting specs and writing code with the help of AI copilots. HR teams are building interview questions and composing offer letters. Marketing teams are generating campaign briefs, customer emails, and press releases. And employees across the average org are using chatbots to summarize meetings, answer policy questions, and even brainstorm new business ideas.
Every one of those interactions leaves behind data—data that’s often proprietary, sensitive, and potentially discoverable.
Why Conversational AI Data Matters
Over 90 percent of Fortune 500 companies already use OpenAI, and adoption is only accelerating. As AI tools become standard across departments, the kinds of data they produce become more varied and high-stakes. Employees are:
- Feeding confidential customer or financial data into LLMs to get strategic recommendations.
- Asking legal- or policy-sensitive questions, such as how to handle internal complaints or contracts.
- Using AI to draft business-critical documents—proposals, emails, policies, and more.
- Uploading files for summarization, editing, or transformation.
- Generating code, content, or images that may raise questions about IP ownership.
Each of these scenarios creates a digital record: a prompt, an output, and context that may be relevant in future litigation, investigations, or audits.
These interactions often include data points like:
- Prompts typed into chatbot interfaces
- Responses generated by the AI
- Files, links, or images shared during the conversation
- Metadata like timestamps, user identities, custom GPTs used, and associated files
The legal implications are clear: AI conversations can surface in matters involving intellectual property, employment law, regulatory compliance, trade secrets, or breach of contract. In many cases, they’ll need to be reviewed, preserved, and produced like any other form of business communication.
Until recently, that was easier said than done. AI chat tools weren’t designed for compliance workflows. They lacked export options, metadata context, or formatting consistency.
That’s why we’ve partnered with leading AI providers to make this data accessible, defensible, and review-ready for legal teams who need to reference it in their Legal Data Intelligence projects.
Starting with ChatGPT Enterprise
In July 2024, we launched our first integration with a conversational AI platform: ChatGPT Enterprise. This was our first step in helping the legal community make sense of this emerging data type.
Our journey into supporting conversational AI evidence started with ChatGPT Enterprise in July 2024. It was a necessary first move; legal teams needed a way to keep up with this fast-emerging data type.
With our integration in Collect, teams can pull full prompt and response histories from ChatGPT Enterprise, along with key context like user details, custom GPTs used, and even images generated by DALL·E, directly into RelativityOne. It’s a connector legal teams need to see the full story—and review it confidently, in a format that actually makes sense.
No guesswork. No screenshots. Just clean, review-ready data.
Next Up: Google Gemini
As generative AI becomes part of everyday work, the data footprint it leaves behind keeps growing. And just as we did with ChatGPT, we’re moving quickly to make sure legal teams can keep pace.
We’re excited to share that RelativityOne now supports collection from Google Gemini, our second major integration for conversational AI evidence in Collect.
This new capability lets you capture prompts, responses, and metadata from Gemini—and bring it into RelativityOne in a clear, easy-to-review format. Like ChatGPT data, it’s automatically converted into Relativity Short Message Format (RSMF), so reviewers can analyze it just like a natural conversation thread.
Rather than being just another integration, we believe this is another step toward making conversational AI evidence as accessible, reviewable, and defensible as any other form of digital communication.
More importantly, the ability to collect Gemini data seamlessly helps legal teams adapt their e-discovery workflows to the rapid evolution of AI tools.
Building a Future-Ready e-Discovery Stack
The speed of change around generative AI is staggering. What started as a productivity tool is now a source of sensitive, substantive content. Increasingly, legal teams must be able to collect and review these records with the same rigor they apply to emails, Slack messages, and Word documents.
Our goal is to help them do just that—by building scalable, secure connectors to the most widely used AI platforms (more to come!), and by translating their output into a format legal teams can work with.
Whether you’re navigating an employment dispute, investigating a potential IP breach, or responding to a product liability claim, RelativityOne gives you the tools to surface and understand the AI-generated content that could shape the outcome——so you stay prepared, compliant, and ahead of the curve as this technology evolves.
What’s Next
We’re not stopping here. As the landscape of AI tools continues to expand—and as these platforms become more embedded in everyday business workflows—the need to account for their output in legal and compliance matters will only grow.
Our mission remains the same: to enable our customers to manage this next generation of data securely, defensibly, and with confidence.
Try It Today
Support for Google Gemini is now live in Collect in RelativityOne. If your organization uses Gemini or ChatGPT, now’s a great time to start building conversational AI into your legal and compliance workflows.
Kazu Shigenobu is a product marketing specialist at Relativity, leading go-to-market strategy and execution for the company’s modern data solutions.
