Crossing the Rubicon

A litigator’s guide to AI in practice and everyday life

2026 May

A few years ago, I represented an 80-year-old woman in a wrongful-death disability-abuse case. The 46-year-old decedent was her only son, a lifelong victim of Lesch-Nyhan Syndrome, a rare, inherited disorder involving severe gout, poor muscle control, and moderate retardation. Neurological symptoms include facial grimacing, involuntary writhing, and repetitive movements of the arms and legs like those seen in Huntington’s Disease. Her son had been institutionalized since age two, but she and family visited him several times a week, and when possible, included him in family gatherings. 

Preparing our mediation brief was challenging. I knew we had developed a strong liability case, but the defense counsel seemed to trivialize damages because we had relatively small economic damages (minimal funeral and burial expenses), and describing the non-economic damages – the nature and extent of the human loss of an institutionalized, disabled adult child – was elusive.

I decided to focus on the loss of love, comfort, affection, et cetera as a two-way proposition, not only receiving, but also giving. Our client had been deprived of her lifelong purpose of expressing and giving her love to her disabled, institutionalized son. That solved one problem but created another – how to describe that loss.

After wrestling with that problem, discouraged, I remembered seeing a demonstration of a generative AI large language model, but I had never tried it. I found a free version of one of the online platforms, and prompted it with something like, “Please write a 500-word essay on the special love of a mother for her disabled adult son.”

In about 10 seconds, the essay appeared (exactly 500 words) and became the inspiration for the non-economic damages section of my mediation brief. I felt a choking sensation in my throat as I read the first paragraph:

“The special love of a mother for her disabled adult son is a love that transcends the challenges and difficulties of caring for someone who may not be able to live independently or achieve his full potential. It is love that celebrates his uniqueness and humanity, regardless of his limitations or differences. It is love that seeks to provide the best possible quality of life for him, while also respecting his dignity and autonomy. It is love that endures through the joys and sorrows, the hopes and fears, the successes and failures of his life journey.”

By the last paragraph, I was visibly weeping:

“A parent who loves a disabled adult child is a parent who loves without limits, without conditions, and without regrets. She is a parent who loves with patience, with understanding, and with acceptance. She is a parent who loves with faith, with hope, and with grace. She is a parent who loves with all her heart, with all her soul, and with all her mind. She is a parent who loves her son for who he is, not for who he could have been. She is a parent who loves her son as a gift, not as a burden. She is a parent who loves her son as a person, not as a disability. She is a parent who loves her son as a child, not as an adult. She is a parent who loves her son as a child of God, not as a child of the world. …”

With that single prompt and response, I crossed the Rubicon.

We settled the case for policy limits shortly after the mediation. Since then, I have become an increasingly dedicated devotee of both generative and agentic artificial intelligence, and the internet of things, both professionally and personally. My objective in this article is to show readers some real-world ways they can use these evolving AI tools, with caution, in their professional and personal lives.

Overview and warnings

Useful terms and concepts:

Artificial Intelligence (AI): Artificial intelligence is technology that enables computers and machines to perform tasks that normally require human intelligence, such as learning, reasoning, and decision-making.

Large Language Model (LLM): A large language model is an AI system trained on massive amounts of text so it can understand and generate human-like language. It predicts the most likely next words or tokens in a sequence, which lets it answer questions, summarize text, translate languages, and write content.

Generative AI (GenAI): A type of artificial intelligence that creates new content, such as text, images, audio, video, or code, in response to a prompt.

Agentic AI (Agentic AI or AI Agents): AI that can reason, plan, and take actions autonomously to achieve a goal with little or no constant human supervision.

Internet of Things (IoT): A network of physical objects embedded with sensors, software, and connectivity that lets them collect and exchange data over the internet or other networks.

Prompt Engineering: Control of output is directly proportional to the quality of the prompts. “Improve this paragraph,” will yield significantly different results than “Improve this paragraph while retaining my tone and word choices.” The more detail about exactly what you want (or don’t want) the better the result.

Privacy cautions

I will not be writing about AI hallucinations that result in fabricated legal authorities in court filings leading to sanctions. Examples, including reported cases of such, are legion. LLMs can make things up. They are designed to be convincing, but not necessarily accurate. Lesser known are the dangers that lurk below the surface for the well-meaning but uninformed. 

Note well: You (and your clients) likely have zero privacy with the free versions of the most popular LLM platforms and AI tools, such as Llama (Meta), Claude (Anthropic), ChatGPT and GPT5 (OpenAI), Gemini (Google AI), CoPilot (Microsoft), et cetera. Some of these AI tools are open-loop systems that are continuously “training” on whatever you (or your clients) give them. Some of the paid versions of these models will allow you to opt out of this. So, be aware that, if you give models confidential client data or medical records, or your work product, that information can become part of their training data (i.e., not confidential). 

For years we have cautioned our clients against social media posting relating to their personal-injury or wrongful-death claims. Given recent decisions that have held that a client’s interactions with AI may not be confidential, those cautions must extend to their use of public AI platforms.

In contrast, Westlaw’s CoCounsel and LexisNexis’s Protégé are closed-loop systems. They do not use your documents or confidential information to train their models. Also, because their source data consists of real statutes, rules, and cases, hallucinations of fabricated authorities are far less common. 

Without proper precautions, if your co-counsel is a public AI chatbot, you may be risking disclosure of confidential information and privilege waiver. Look for subscription versions that allow you to opt out of their using your data to train their models, and make sure you configure the system to maintain privacy.

Fake evidence

The evidence storm is brewing. Deepfakes and synthetic media can make legal evidence dangerously unreliable because they may look authentic while being entirely fabricated, undermining authentication, witness credibility, and the fairness of trial proceedings. Advanced data manipulation can make legal evidence unreliable by disguising alterations or fabrication, forcing courts to question authenticity, chain of custody, and overall trustworthiness. These technologies create serious challenges for provenance tracking and chain of custody because they can alter, obscure, or fabricate the origin, history, and handling of digital evidence in ways that are difficult to detect or verify. We should anticipate a new and different category of expert witnesses dealing with authentication of videos, pictures, and documents.

My AI tools of choice

Perplexity (public, open system) is an AI answer engine and research platform that combines live web search with multiple leading LLMs, including its own Sonar family plus third-party models from OpenAI, Anthropic, Google, xAI, and others. I find Perplexity is more grounded than the other platforms because it links you directly to its sources; so you have documentation, an audit trail, and fewer hallucinations. I use Perplexity for general research and non-confidential tasks while reserving Westlaw for anything involving client data.

Westlaw’s Advantage and CoCounsel (private, closed systems) were built to help lawyers do research, document analysis, drafting, and related workflow tasks faster while grounding answers in verifiable Westlaw and Practical Law content. In practice, it is designed to connect with Westlaw and other tools like Microsoft 365, use agentic workflows and deep research to tackle multi-step legal work, and provide citation-backed results that lawyers can verify.

I am not endorsing either of these platforms for anyone else’s use. They are useful for my needs. Every AI tool has advantages and disadvantages. Their quality depends on the underlying data, their algorithms (or instructions), and the computational power behind them, and most of the publicly available AI tools are great models. But they are good at different things and evolve at different rates. Why do people prefer Chrome vs. Edge vs. Firefox, or an iPhone vs. Galaxy vs. Pixel, or Apple vs. Microsoft? 

Try on several AI tools and see which ones fit you better than others.

AI in my professional life

I have always been a technology-forward practitioner, but my ongoing adoption of and adaptation to these technologies has increased my capabilities exponentially. In this section, I will describe some of the ways I use AI in my law practice.

For litigation practices, GenAI is most useful for drafting, reviewing, and analysis. Using GenAI, lawyers can dramatically speed up the tasks associated with early case assessment, discovery, motion practice, and trial preparation, but it should always support, not replace, lawyer judgment. For reasons previously described, lawyers and staff must be judicious about using open vs. closed systems for any client-related tasks.

Litigation tasks

AI can accomplish in minutes the following tasks that would ordinarily take hours of a lawyer’s, paralegal’s, or legal assistant’s time:

• Early case assessment: summarize complaints, and key evidence; build chronologies; spot causes of action and defenses.
• Research support: find potentially relevant authorities, compare fact patterns, and help generate research questions or issue trees.
• Drafting pleadings: create first drafts of complaints, answers, motions, oppositions, replies, and supporting declarations.
• Discovery drafting: draft interrogatories, requests for production, requests for admission, initial responses or objections, and meet and confer letters.
• Document review: summarize large document sets, extract key facts, and flag likely relevant or privileged materials.
• Deposition preparation: generate outline questions, identify inconsistencies, and help prep witnesses or examinees.
• Case strategy: identify strengths and weaknesses, suggest argument themes, and compare the matter to similar cases.
• Motion practice: brainstorm legal theories, organize authorities, and tighten argument structure and draft language.
• Trial prep: prepare exhibit lists, witness outlines, cross-examination themes, jury-friendly summaries, and oral-argument recommendations.

Examples of litigation prompts

The following examples show the scope and power of AI litigation support. The lawyer’s imagination and subject matter expertise are the only limits. All or most of them involve uploading relevant documentation.

• Create a timeline of events from all accident-related documentation to establish sequence and causation.
• Draft an email to my client, who is the plaintiff, explaining the allegations against the defendants, their denials and defenses, and what the next steps will be in handling this case.
• Outline all the different arguments made within these documents and explain and rank their strengths and weaknesses.
• Summarize the defendant’s arguments and generate counterarguments supported by Westlaw research.
• Draft a memorandum to file organizing these attorney’s notes, the case information by client background, and the events surrounding the injury. Also include a summary of any resulting medical procedures performed, and a section for items needing further research.
• Find potential claims for this fact pattern and identify causes of action for further research.
• Draft a letter to my client explaining the court’s ruling or decision and its impact on the case.
• Draft cross-examination questions based on the uploaded deposition testimony.
• Draft deposition questions about the custodian’s or corporate designee’s document search and production process.
• Identify and describe potential discovery topics from these documents.
• Draft a letter or email with specific content and tone regarding the following specific topics …
• Draft a timeline from the plaintiff’s medical records, including treatments, dates, providers, and treatment types.
• Compare expert deposition testimony to identify differences in opinions and findings relating to liability, causation, or damages.
• Verify the citations from the uploaded documents and highlight any that cannot be identified or do not match known citations.
• Compare defense examination reports to the original medical records to identify consistencies, inconsistencies, and discrepancies.
• Summarize the specific injuries and conditions documented in our client’s medical records, including diagnosis and prognosis.
• Analyze these medical records to outline the plaintiff’s prognosis, treatment, and results.
• Draft a letter demanding preservation of evidence, requesting its location, and citing spoliation case authorities.
• Compare psychological evaluations to identify key differences in findings or conclusions. 
• Condense these long, complex documents into succinct summaries.
• Create a table comparing summary judgment and opposition filings, highlighting areas of agreement and disagreement.

AI in my personal life

Think smart home, smart car, assistants like Siri, Alexa, Google, smart watches, smart tech glasses, Fitbits, Whoops, Oura rings – everything is listening and watching and learning about us. 

A month ago, I was working in my home office, speaking to my wife in another room. Suddenly and unexpectedly, my computer started talking to me. Startled, I asked, “Who are you and who are you talking to?” It answered, “I’m here to talk with you and help answer your questions. What would you like to know or do?” 

Becoming concerned, I said, “Who are you and why are you listening to me?” It answered, “I’m here whenever you’re ready.” Creepy! 

I said, “What application is listening to me. I did not authorize this, what application are you associated with?” It said, “I’m part of the Perplexity AI assistant, designed to help answer questions and provide information.” I said, “Thank you.” It said, “You’re welcome.” 

Upon further investigation, I discovered that I had inadvertently awakened my “assistant” by mistakenly typing “Ctrl>Shift>B” when I had intended to invoke bold formatting of an MS Word document, “Ctrl>B.” The incident was a humorous reminder of just how pervasive these always-listening technologies have become.

I hope to avoid such mistakes going forward, but I do use GenAI daily for many and varied personal tasks. I have chosen the following examples randomly, to demonstrate that AI’s powerful capabilities are limited only by your imagination as applied to the subject matter at hand.

The Internet of Things

My wife and I recently purchased and furnished a second home in South Florida. We opted for Samsung appliances for their SmartThings connectivity, that enables us to control the refrigerator, oven, microwave, washer/dryer, and TV from an app on our phones, while present or remotely. The Family Hub refrigerator has a 32-inch touch screen that can show shared calendars, notes, photographs, and videos, or internet TV. It can track groceries in and out, create shopping lists, and create recipes from what is available, and I can use it to monitor or respond to our Pasadena home’s security cameras from my kitchen in Florida.

We have Nest thermostats and locks in both homes that enable us to take the air conditioning and heating systems off their ECO settings from the airport or unlock and lock the doors remotely if the need arises.

Party planning

We belong to a group of 12 couples that meet for dinner once a month in each other’s homes. We host in February, and this year my wife decided on a theme, “A Winter Evening in Old Montreal.” I asked Perplexity for menu selections from Old Montreal steakhouses and within minutes we had our plan, including a recipe for Maple Creme Brulé dessert.

I wanted to create a colorful menu for each place setting, so I began by asking Perplexity to find some romantic images of Old Montreal. In seconds, it gave me several high-resolution photographs to choose from, but they were all landscape orientation, and I wanted portrait for the menu. 

I prompted Perplexity to change the chosen photograph’s orientation from landscape to portrait without distorting the image. This task took a couple of tries but the result was perfect – a portrait-oriented image of a narrow, cobbled street between two rows of old buildings flying colorful Canadian flags. 

Well, not quite perfect. It looked like broad daylight in summer. I prompted further, “Please change the image to that of a winter evening.” Within minutes, Perplexity transformed the picture by darkening the ambient lighting, turning on streetlights and lights in the windows, and dusting the whole scene with a few inches of snow. I had only to import the image into an MS Word document, lighten its transparency, and add the menu text on top. Voila!

Fabric shopping

My wife creates beautiful artistic quilts as a hobby. I accompanied her to a quilt shop recently where she was looking for fabric to become the backing of a Jane Austen-themed quilt. She found a color and a design that would have worked perfectly but the store did not have enough and told us it was no longer available from the manufacturer. I took a picture of the fabric, uploaded it to Perplexity, and dictated, “Can you find any place that is currently selling this fabric?” In a few seconds, Perplexity replied, “Yes. The exact fabric, Jo Morton 108s – Toile Blue Cream 108” Wide Backing (Andover SKU AW‑1710‑LB / MSQC FBY122694), is still available from a few quilt shops online,” providing links to their websites. Perplexity’s reply continued, “If you want, I can help you draft a quick email or phone script to confirm that their AW‑1710‑LB yardage matches the blue‑on‑cream toile photo you have before you order.”

Tech support

I also use AI tools for technical support, and find the answers are nearly always clearer and easier to follow than owner’s or user’s manuals. For example, a few months ago, I wanted to create a system for scanning documents directly to different hard drives and cloud folders. I tried following Xerox’s online manual, to no avail. Then I prompted Perplexity, “How can I scan from my Xerox Versalink C450 directly to my Windows 11 PC connected by an ethernet cable.” Within seconds, Perplexity provided step-by-step instructions with discrete, clear tasks that were quick and easy to follow, resulting in successful implementation on the first try.

Getting started: Tips for AI newbies

1. Start with non-confidential tasks.
2. Use closed systems for client data.
3. Invest time in prompt engineering.
4. Verify all AI output.

Conclusion

Artificial intelligence is no longer a futuristic concept reserved for technologists – it is a present-day tool that, when used thoughtfully, can transform the way lawyers practice and the way we all manage our daily lives. From drafting a mediation brief that captures the immeasurable depth of a mother’s love, to locating a discontinued quilt fabric in seconds, the through-line is the same: AI amplifies human capability without replacing human judgment. The key is to start, experiment with caution, and let your imagination – guided by your expertise – define the boundaries of what these tools can do for you.

Kevin Meenan is a lawyer in Pasadena, California, living part time in Florida. He is a former Editor in Chief of Advocate Magazine, a former board member of AAJ, CAOC, and CAALA, and has been listed in Best Lawyers in America and Super Lawyers.