Finding Value in an AI World
The value comes from you.
If you’ve looked at LinkedIn in the past year and a half to two years, pretty much all of it seems to be people posting something like this:
AI will replace lawyers who use AI to replace lawyers who don’t use AI to replace clients who use AI to replace lawyers who don’t use AI, who will then replace clients who don’t use AI to replace lawyers who use AI. Here’s why that matters:
[a picture of poster looking at the camera]
I don’t necessarily disagree that AI is creating some type of seismic shift, but sometimes it feels like we’re all shouting about doing stuff without understanding what it is we’re doing, or why we’re all shouting.
A few days ago Google announced Gemini Enterprise, a potential competitor to Microsoft’s Copilot. What was more interesting to me in watching the video announcement was this:
A year ago, Google’s AI services were processing around 65 trillion tokens per month.
Right now they’re processing around 1.3 quadrillion tokens per month.
Let’s not leave out OpenAI - they announced at their Dev Day that they’re processing 6 billion tokens per minute through their API.
At this point, exponential progress seems inexorable:
Think about this: in early 2023 lawyers everywhere (including me) lambasted the people behind Do Not Pay for offering a $1 million bounty to anyone who let their AI tool (a wrapper on ChatGPT 3.5 [shudders]) argue their case before the U.S. Supreme Court. Do Not Pay then pivoted to a plan to use it in California traffic court, which was ultimately shut down by the California Bar’s threats to prosecute them for the unlicensed practice of law.
Fast forward to 2025, where a lawyer who argued a real case in front of the U.S. Supreme Court and won decided to see how AI would do with it. His conclusion?
Was it better than me? Well, you all can decide that. But:
All of Claude’s answers were clear, coherent, and directly responsive.
Claude skillfully batted hostile questions away and used them to bolster its core themes.
Claude never stumbled or got tongue-twisted.
Claude knew the record cold.
Claude gave several unusually clever answers, making arguments I didn’t think of.
…
In my opinion, AI will soon surpass even the best human oral advocates, if it hasn’t already.
Let’s add to that: real people are using AI to replace lawyers in court, and are winning. Keep in mind that these are people who probably couldn’t afford a lawyer in the first place:
“It was like having God up there responding to my questions,” White said after using the chatbot and an AI-powered search engine called Perplexity to help represent herself in court.
…
White regularly provided ChatGPT with documents and detailed information about her case. The chatbot helped White identify potential errors in a judge’s procedural decisions, chart out possible courses of action, research applicable laws and draft responses to the court.
Several months of litigation later, White managed to overturn her eviction notice and avoid roughly $55,000 in penalties and more than $18,000 in overdue rent.
This isn’t to say that just workslopping stuff into a court filing will work - AI hallucinations still happen and will get even a pro se litigant sanctioned. But at this point the progress seems either inevitable or inexorable. Maybe both.
Do tokens = value?
Consider: Google’s 1.3 quadrillion tokens per month, or OpenAI’s 6 billion tokens per minute - that’s a metric shit-ton of … stuff? Are we seeing some type of value from that? At some level it feels like we’re too all busy trying to add AI to processes that may or may not matter, instead of going through the hard process of figuring out what actually matters. Suggested LinkedIn post: “Here’s what that taught me about B2B sales.”
In the meantime, the AI products space feels like an endless carousel of things that can put in one piece of a puzzle but don’t integrate well with anything else. It’s kind of like seeing one of my many pet peeves with science fiction - robots sitting in chairs and typing on computer consoles - come to life. Meanwhile OpenAI, Anthropic, and Google are busy creating integrations that work across so much stuff it’s mind-blowing.
Speaking of robots typing things into computers, the kind of AI we have now isn’t contemplated in most of popular science fiction. In most sci-fi, AI is either embodied or contained in a physical form (R2D2, Cmdr. Data, Wall-E, the Cylons) or is an all-powerful and pervasive intelligence single-mindedly seeking human destruction (Skynet, the Borg, Microsoft Teams1). What we have now is almost like a Domovoy, or house spirit, in every product we use. Or maybe it’s a golem that can speak and doesn’t always follow directions literally. Maybe we don’t have good metaphors for this.
Integrations still matter:
While I was working on this post yesterday, Clio announced “Clio Work” - which looks like a completely integrated AI solution that does end-to-end work inside the Clio platform, intake to invoice. Considering that it, as I understand it, will also combine document analysis and legal research with vLex’s Vincent, and document drafting with Clio Draft (I think it’s called Draft), it could be the single entry point for legal practice. For years Clio has described its product as the “operating system for law” and this is (potentially) that.
We’re also getting close to a point where one could hook up ChatGPT to Sharepoint, Google Drive, Outlook, Hubspot, or whatever and just do work by typing into the magic box. Or even talking to it. It may not be long before we move from AI assistants in each discrete tool2 to having ChatGPT, Gemini, or Anthropic be the default gateway to doing work.
In under 3 years, we’ve gone from a lawyer on Mata v. Avianca clicking through the “this is an experiment” warning on ChatGPT 3.5 to find fake case law, to AI being almost able to grind through analysis, research, and the drudgery of firm operations in the same browser window. This isn’t to say lawyers should wire up their Sharepoint to ChatGPT! Don’t do that if you don’t understand the privacy and security implications!
How to figure out the problem of value:
It could be that “deploying AI” is as easy as buying a product or following the three steps one of the marketing emails we all get. I’m convinced that finding value is much, much more difficult.
Here’s an idea - value doesn’t come from the discrete task or the tool itself. Think about a carpenter: does the value in what the carpenter does come from the tools they use? Or does it come from the beauty and functionality of the finished product? The owner of a new chair or bookcase doesn’t see the end result of carpentry as a bundle of cuts, wood, screws, joints, and glue. Likewise your clients don’t see your work as a bundle of Outlook events, Redlines, Word templates, time entries, or searches in Wexis. Don’t lose sight of that.
I wrote a little while ago about why it’s tempting (but wrong) to see the role of lawyer as bundle of tasks. Were lawyers lawyers because they researched in books? Does researching cases on the internet make you less of a lawyer? What if there was an AI tool that completely eliminated Bluebooking - 100% accurate, 100% of the time. Does that lessen the importance of lawyers to their clients? To the legal system?
I know I’ve talked a lot about finding value but haven’t given you “Sam’s Three Rules for Finding Value” or whatever. Sorry not sorry. Try this instead:
Call up one of your past clients, hopefully one that you trust, and who likes you;
If you bill by the hour, tell them this is not something you’re going to charge for.
Ask them something like this - “How do you see the value of what I did for you?”
Or: “Can you tell me, in your own words, how I helped you?”
Write it down. Call another client.
I don’t think your clients will tell you that you had great templates, did the best Bluebooking, or deployed the most effective prompts or efficient workflow automations. Hopefully they tell you something like “you made a difficult process simple” or “you were there to guide me.” Does AI get you closer to doing more of that?
The biggest thing that bugs me about the “Lawyers using AI replacing …” trope is that it treats AI like some kind of magic bullet. These tools seem magical but that doesn’t make them magic. I really believe AI can help you enhance the value you give clients. But without understanding the value you give your clients, you risk adding something that hinders, not helps. Value doesn’t come from an AI tool. The value comes from you.
I’m kidding! Definitely kidding!
Another pet peeve of mine - Jehoshaphat H. Christmas try improving your tool’s usability before you add an AI assistant.