How Anthropic uses Claude in Legal

1. MPMark Pike0:00

   [gentle music] This is a legal lamp is what we call it, and I was working on a project to learn more about how Claude Code works. And so I was trying to think of a fun project that would sort of make my desk come to life more. And now I can type a message and ask this lamp to blink in Morse code. It says it's thinking. It's gonna do four quick blinks and two quick blinks, which is H-I. Abracadabra. It's kind of magical. I'm not an engineer. I'm non-technical. I don't know how to code, but with Claude Code, you don't have to know how to code. Uh, you can just kind of talk to it in plain language. No lawyer likes doing the same repetitive exercise over and over and over again. The, the work can be dull. You can make mistakes. Before Claude, I had a ton of busy work, things I would put off to the end of the day 'cause I just knew it would take a lot of time, but not using the best parts of my brain. So we put our heads together and realized we might be able to use Claude to help get our best work done and, and build workflows. It used to be that the, uh, marketing team would reach out to me maybe a day before a launch and say, "Hey, we've got this really exciting launch happening tomorrow. We're so sorry. The blog post just came together. We need you to quickly read this and flag anything that might be problematic." So, uh, I asked Claude once to just build me a workflow. Here's what I care about. And to my surprise, Claude just ran with it and, and set it up. If I were a marketer and I needed to get one of my blog posts reviewed, I would open up this link to the marketing material self-review tool, which is pinned in my channel. I would cut and paste. So let me copy their blog post. I'm gonna go back to the review tool. I'm gonna paste all that content, and then I'm gonna click the Analyze Content button, and this is going to send Claude off on its journey reviewing my material. There we have it, review results. Claude has identified five issues to address, so it wants to make sure I focus on accuracy. It wants to review the security claims, wants to make sure we have publicity rights on third-party content, as well as partnership considerations. And it gave me props. It said what looks good, clear the feature descriptions, comprehensive integration details, specific use cases. Here are the items that require legal ticket review. I'm gonna click on Generate Slack message for legal team, and Claude has now summarized the issues here. I'm gonna click on Copy to clipboard, and I'm gonna click Go to legal tickets to file my ticket to tee it up for legal review. It, uh, identifies the most important issues. It helps me prioritize, sort of has like a low, medium, high risk level signal, uh, that's based on a framework that I gave it. So it's kind of acting as my eyes and ears for that first pass. As we build these types of automations and, and new workflows, I, I'm always trying to ensure that a, a human is, remains in the loop from the legal team. We know that AI systems can still hallucinate. I'm still making sure I'm reviewing the work, uh, but this is really helping us move with more speed and sort of preemptively flagging things. Within the legal department, we're using Claude in so many different ways. So we're using it to do redlining exercises in commercial, uh, matters. We're using Claude to help us, uh, with our conflict of interest policy and reviewing outside business activity requests. When people reach out to me and ask how do I get started, uh, I tell them to think of their most routine work. Just open up Claude and, and give it a shot. You really don't know what it's capable of doing until you give it a shot. [gentle music]

Episode duration: 3:39