How can GenAI help you start assessing the allegations in a case right away? With this allegation extraction prompt from Aaron Patton!

In a LinkedIn post this morning, Aaron (aka Gates Dogfish) published this post on LinkedIn, where he said: “Attorneys who haven’t logged into an eDiscovery platform in a decade can fire up Chrome and start meaningful ECA on day one of a matter.”

Where do you start? Aaron says this:

Here’s what I would do:

1. Upload the complaint to ChatGPT/Gemini/Claude (most are public documents).

2. Use the Allegation Extraction prompt below.

3. Review the questions generated: tweak or discard them as appropriate.

4. Paste a question into your GenAI/RAG tool.

5. Read the answer, copy it to the table, and binder the docs to save them for review/assignment/TAR/coding.

6. Repeat 4 & 5 for every question.

Here’s his Allegation Extraction Prompt:

“You are an eDiscovery analyst and search expert. Given the full text of the attached legal complaint, extract each distinct factual allegation. For each allegation, write a simple, neutral, evidence-seeking question that could be used to query ESI hosted on a GenAI/RAG-powered discovery platform. Avoid compound or timeline-based questions. Focus on “who,” “what,” “when,” “where,” “why,” or “how” formats.
Output Format: A two-column table with:
1. Allegation
2. Proposed Question”

I decided to perform the first two steps to see what allegations I could get out of a reasonably large complaint using GPT-5. The case I picked was ironic – The New York Times complaint filing against Microsoft and OpenAI in December 2023 (complaint available here). 69 pages, so it would take quite a while to get through manually.

GPT-5’s Thinking mode “thought” for 2 minutes, 55 seconds and then it delivered 34 separate allegations from the complaint and a proposed question to go with each.

Here’s what the prompt and the first part of the response looks like in GPT-5:

The links to “NYT Complaint Dec2023” simply download the PDF of the complaint that I uploaded – which makes them useless (in this instance, at least). After the results, GPT-5 added:

“If you want this exported to CSV/Excel or refined for specific custodians, date ranges, or repositories (e.g., Azure logs, Bing index artifacts, training data manifests, product telemetry), say the word and I’ll format it for direct upload into your discovery workflow.”

So, I asked GPT-5 to export the results to CSV format. It did, and I formatted the results, saved it as an Excel and also generated a PDF. The full response to the prompt is in that file here.

Did I check the allegations to make sure they were there? I tried, but I did a search on the first one in Acrobat and it was taking forever, so I canceled the search. Will try again later – in the meantime, would welcome any feedback on the results.

Regardless, this Allegation Extraction Prompt is a terrific example of how you can leverage GenAI tools to get a jumpstart on preparing your litigation response and it shows you just how powerful these tools are. Thanks, Aaron!

So, what do you think? How is your organization using generative AI to prepare for response to litigation? Please share any comments you might have or if you’d like to know more about a particular topic.

Image created using Microsoft Designer, using the term “robot lawyer looking at a litigation complaint on a computer”.

Disclaimer: The views represented herein are exclusively the views of the author, and do not necessarily represent the views held by my employer, my partners or my clients. eDiscovery Today is made available solely for educational purposes to provide general information about general eDiscovery principles and not to provide specific legal advice applicable to any particular circumstance. eDiscovery Today should not be used as a substitute for competent legal advice from a lawyer you have retained and who has agreed to represent you.