The Best (and Worst) Use Cases for AI in Litigation

There's lots of discussion about law firms adopting AI in the abstract, but you rarely hear about how litigators actually use it day to day. Having spent eight years as a litigator at Clifford Chance, WilmerHale, and Willkie Farr & Gallagher before founding Crimson, I've seen both the promise and the limitations of AI in practice.

Here are three of my favourite use cases for AI in litigation, plus three tasks I still wouldn't use LLMs for.

Great Litigation Use Cases

Case File Analysis

Litigators spend most of their day dealing with lengthy documents, and LLMs are ideally suited for text-based analysis. The applications here are enormous:

• Mapping party positions across pleadings and correspondence
• Comparing witness evidence to identify consistencies and contradictions
• Reviewing hearing transcripts for admissions or problematic statements
• Summarising judgments and extracting the key holdings
• Digesting technical expert reports and explaining complex concepts

The right AI can save disputes teams countless hours on these tasks while actually improving the quality of the work product. When you can analyse every document rather than just a sample, you build a more complete picture of the case.

Correspondence Management

On complex cases, handling inter partes correspondence can be a full-time job. Someone needs to track what's been said, identify deadlines, and ensure nothing falls through the cracks.

Litigation-focused AI is excellent at correspondence-related tasks:

• Summarising and tagging letters and emails
• Creating chronologies of key exchanges
• Finding specific quotes and admissions
• Drafting procedural letters in a firm's house style
• Tracking requests and responses across lengthy correspondence chains

It's unglamorous but essential work – and AI handles it remarkably well.

AI as Supervisor and Sparring Partner

One underdiscussed benefit of AI is how patient it is. You can ask it anything, and it won't judge you.

• How do constructive trusts work again?
• What do the CPR say about without-notice applications?
• Are there any points I didn't explain well enough in this draft submission?
• What would opposing counsel argue in response to this point?

With access to the right sources and proper safeguards to prevent hallucinations, AI can help lawyers become better litigators much faster than before. It's like having a knowledgeable colleague available 24/7 who never gets impatient with basic questions.

Not-So-Great Use Cases

Outcome Prediction

Can AI analyse pre-action exchanges or pleadings and identify which side will prevail at trial? This is an area where I'd urge caution.

For high-volume claims that are factually similar – certain types of medical negligence or personal injury cases, for example – AI may well identify useful patterns in historical outcomes. But complex commercial disputes are different. They often turn on:

• Heavily contested witness evidence
• Protracted disclosure battles
• Novel legal arguments
• Credibility findings by the tribunal

I'm sceptical that AI can make reliable predictions in these cases. The variables are too numerous and too qualitative.

Calculations

Most legal AI tools are built to process text, not numbers. In general, LLMs are not great at maths, so I wouldn't use them to:

• Prepare damages figures
• Calculate interest
• Add up time sheets
• Perform financial analysis

For anything involving calculations, use purpose-built tools or do the maths yourself.

Client Emails

As AI adoption increases, I suspect the human factor will become even more important in legal practice. A client's choice of lawyer often comes down to the strength of personal relationships – trust, responsiveness, and the sense that their lawyer truly understands their situation.

I wouldn't rely on AI to produce finished client emails. Use it to help structure your thoughts or gather information, by all means. But the final communication should be authentically yours.

The Bottom Line

AI is a powerful tool, but it's not magic. Understanding where it excels and where it falls short is essential for getting real value from legal AI. The litigators who will benefit most are those who deploy AI thoughtfully, leveraging its strengths while maintaining human judgment where it matters most.