.jpg)
At five o'clock the evening before a motion to compel hearing, opposing counsel filed a surreply. Federal case law I had not briefed, in a state matter no less, a 244-line privilege log, the hearing set for nine the next morning. It was built to leave me up all night and underprepared, or crawling to the judge for a continuance.
I walked in the next morning and tore them up. They thought they had caught us flat. I had it figured out before I went to bed.
The way I got there is not what people picture when a lawyer says he used AI. I did not paste the filing into a chatbot and read back the answer. I loaded their cases and the privilege log into Eve, set them against the discovery we already had, and pushed: distinguish every case they cited under Oklahoma law, flag the log entries that never should have been on it. By the time I closed the laptop I had my oral argument and a list showing most of the log was junk, documents created before the incident, nowhere near privileged.
The machine did not win that hearing. My questions did. Point a tool at your records, read back whatever it hands you, and you walk into court confident about things you never checked. I make it prove every answer before I rely on it.
It did not make me a better lawyer than I was the day before. It gave me back the hours.
Eve has read the entire file. It can find any document, surface any inconsistency in a deposition, faster than a room full of young lawyers.
Which of those things matters is your call.
I was working up a wrongful death matter built on roadway construction, a thicket of contracts and competing standards. I could talk to the record like a colleague, ask how a theory held up, and it would walk me to the documents that carried it. But it never once told me which theory a jury in a death case would actually reach for. That call came out of twenty-two years of trying these cases, and a focus group that told me which version they believed.
The tool is the brilliant associate who has memorized every page and never sat through a voir dire. You still bring the questions, and a lazy question buys a confident, worthless answer.
The first question I put to every young lawyer on a new case is the one they get tired of hearing: how do we lose this? People assume a plaintiff's lawyer wants a machine that tells him he is right. I want the opposite.
So the first thing I do with a serious case is load it into Eve and tell it to play defense. Give me their theory. Find the ten worst facts in the file, and show me where my own timeline goes soft. By the afternoon I have the argument I used to wait until the reply brief to see, early enough that I can still fix what needs fixing.
The value has nothing to do with the computer being clever. It is tireless, and it feels no loyalty to my case, so it will lay out exactly how I lose if I tell it to. I tell it to every time.
The first demand letter I had it draft was a problem, and not the one people warn you about. It was too complete. It laid my whole theory on the table before I had made the other side work for a single piece of it.
A plaintiff's lawyer can be too specific. You can hand the defense a map of your case months before you want them to have it.
So I treat what it gives me as a draft with the volume turned up too high, and my job is to bring it back down. I do the same with discovery: better to have it build something fresh for the case in front of me and edit hard than fire the same stale form at every file.
The work now is knowing which words to take back out.
I will not put a citation in front of a judge because a computer told me it was good. I pull the case, I read it, I confirm it says what the machine says. We have all watched lawyers get burned trusting bad quotes from other systems. The one who lets that rule slip is going to have a very bad morning in a courtroom someday.
The quieter danger is that you start to lean. Before a deposition I still go through the file by hand, the way I did for decades, because that is how I find the questions I did not know were there. Let Eve pull the page and line where one witness contradicts another, but the preparation stays mine. It is a tool on the belt, not the hand that swings the hammer.
That is what worries me for the lawyers behind me. A young attorney who skips the reading never builds the instinct that tells you, three years in, which document is the whole case. He is renting a skill instead of owning it, and the record exposes him at the worst possible moment.
This is the most powerful tool I have seen since Westlaw replaced the stacks, and I do not say that lightly. It has not made me any less obsessive about a case; I still look at the same document a hundred times and take my own photographs at the scene. Use it to argue against yourself. Use it to do in an hour what used to eat a week. Then go read the file, because the case was always going to come down to the lawyer and not the machine.
Jeremy Thurman is a partner and trial lawyer at McIntyre Law in Oklahoma City, where he handles catastrophic injury and wrongful death cases.
