When Mike Whelan invited me onto his new podcast, AI in Practice (presented by Spellbook), I expected we’d discuss legal research tools and workflow automation. We covered those topics, but most of our conversation centered on a bigger question: what expertise means when AI can analyze a 5,000-page record in minutes.

My consultative practice model

I run a solo appellate practice supported by a small team—a paralegal, a controller, and an administrator—so I can focus on issue selection, strategy, narrative, and advocacy.

Trial lawyers across Texas bring me in to:

  • Embed with their teams and provide guidance in real time.
  • Appear for important hearings and motions.
  • Preserve appellate issues by building a clean record before, during, and after trial.
  • Handle any appeal through resolution.

This model works because I focus on high-level tasks—spotting issues, developing strategy, crafting persuasive narrative, and advocating based on three decades of civil appellate experience. As a solo, I need ways to pressure-test theories, expose blind spots, and explore “what if” scenarios.

AI—mainly Westlaw Advantage and Thomson Reuters CoCounsel—has become the thought partner that helps me do this work better and faster. Other tools offer similar capabilities.

Where AI fits the model

AI gives me another set of eyes on the record, helps me generate and challenge arguments quickly, and expands my capacity without adding headcount. It’s like having a tireless second chair who accelerates my analysis while I retain the judgment and accountability for what goes into the brief and what gets left out.

Here’s how it helps:

  • Research assistance. Deep-research tools let me move from issue framing to vetted authorities fast. AI accelerates the search and screening; I decide which cases carry the narrative and which to exclude.
  • Closed-data-set analysis. AI can quickly process transcripts, records, and briefs for pattern recognition and analysis. During a lengthy trial earlier this year, I used CoCounsel to query daily transcripts, which allowed me to identify which witness addressed a specific issue and get an answer with highlighted excerpts in seconds.
  • Argument development. Westlaw can analyze opposing briefs and suggest counterarguments, which is a starting point for deeper analysis.
  • Briefing. I can feed an opponent’s brief into Westlaw and ask it to critique the legal reasoning and verify the authorities cited, then evaluate which critiques are valid and which are dead ends.

AI can’t predict a court’s response to a specific argument in a given fact pattern. It can’t account for judicial tendencies, case strength, panel composition, or the dozens of variables affecting appellate outcomes. The calls that win appeals—deciding what is likely to land in a brief or at the lectern and how to frame it—remain mine.

Avoiding the hallucination trap

We’ve all seen sanctions for fake case citations. Hallucinations remain a problem in AI-powered legal work, but they most often result from asking too much of a non-legal-specific tool (like ChatGPT) and failing to verify its output.

I don’t ask AI to draft a brief from scratch, and I critically evaluate its output. That’s no different than how I would review the work of an eager but inexperienced associate. In either case, my name is in the signature block, and the product is my responsibility. AI assists my work; it doesn’t replace the careful review and judgment that competent representation requires.

Circling back to expertise

Mike’s “hobby horse” is that expertise is methodology, not credential. You don’t own expertise because you have the books. You demonstrate it by seeing patterns, extracting insights, and knowing what to do with them. Expertise is method—judgment, persuasion, narrative control, and reading the room—all human work.

I’m the one deciding what goes into a brief. During oral argument, I’m the one at the lectern. The AI isn’t. The court doesn’t decide based on how many cases the AI found; it responds to how I frame the issue, answer questions, and maintain the narrative under pressure.

That’s the difference between AI assistance and human expertise in my practice: the technology helps me see the map faster, but my judgment still determines the route.

If you’re working on an appeal or trial that could benefit from strategic consultation or want to discuss how AI fits your practice, reach out here or connect with me on LinkedIn.