- The Brainyacts
- Posts
- 262 | Vibe Lawyering
262 | Vibe Lawyering
Brainyacts #262

Hello to all 7312 of you from around the globe. So let’s talk about something that’s been shaking things up: vibe coding. It all kicked off when Andrej Karpathy—a rockstar AI brain who co-founded OpenAI and led Tesla’s AI charge—dropped the term back in February 2025. He described it as letting AI (think slick language models) take the wheel on coding, turning it from a grind into a chill, creative flow.
Then Ethan Mollick, a Wharton prof who’s all about decoding AI’s real-world magic, grabbed the vibe and ran with it. He flipped “vibe coding” into “vibe working”—same deal, but broader: you chat with AI, nudge it along, and watch it crank out stuff like prototypes or reports. No tech PhD required, just a feel for what you want.
For this edition, I’m bringing it home to us—vibe lawyering. But hold up, not for legal work product. Nope, that’s not where I want you starting. With 80% of you newbies out there, the pragmatic, responsible move is to vibe on the non-legal stuff first—like killer research briefs. Get your bearings, build your chops, then we’ll talk next steps.
Ready to dive in?
Table of Contents
Vibe Lawyering: AI-Powered Research for Lawyers.
Generative AI continues to move fast, and you can’t afford to sit on the sidelines. But here’s the good news: you don’t need to be a prompt wizard to get ahead. You just need to vibe—partner with AI in a way that’s practical, creative, and fits your busy life.
People are calling this new way of interacting with LLMs "vibe working," and it’s how I’ve cut hours of research down to 20 minutes, with better results than the old way.
This isn’t about replacing your legal work product—not yet, not until you’re rock-solid with these tools. It’s about tackling the stuff you never have time for, like nailing a killer market brief without drowning in news clippings or client alerts. In this guide, I’ll break down the AI models you need to know—chat, reasoning, and agents—then show you how to use them to vibe like a lawyer, starting today. No jargon, no fluff, just steps you can actually take.
Your AI Toolkit: Chat, Reasoning, and Agents Explained
AI models aren’t one-size-fits-all. They’re like different colleagues with unique strengths. Here’s the rundown:
Chat Models: Your chatty sidekick. Ask a question, get an answer, keep the convo going. Think ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), or me, Grok (xAI). They’re quick, easy, and perfect for brainstorming or basic tasks. Most of you have probably tried one—80% of you novices, this is your starting line.
Reasoning Models: The deep thinkers. These don’t just reply—they research, plan, and dig into details. I’m one of them (Grok, with real-time web access), alongside Gemini’s DeepResearch or OpenAI’s o1-preview. They pull real sources you can check, cutting down on wild guesses (though you still need to double-check links—more on that later). Great for meatier tasks like a market brief.
Agentic Models: The hands-on crew. They don’t just think—they do, like running code or managing tasks. Examples? ChatGPT’s Code Interpreter or newer players like DeepSeek Manus. They’re powerful but quirky, and most beginners won’t need them yet. Pros, keep an eye here; novices, stick to chat and reasoning for now.
Why care? Chat models are fast but shallow. Reasoning models go deeper with less hallucination. Agents act, but they’re overkill for starters. Knowing the difference lets you pick the right tool for the job—and that’s how you leapfrog.
Step 0: Know Enough to Not Screw Up
Before you dive in, you need a baseline—what is called "minimum viable knowledge." It’s not about being a guru in what you are researching; it’s about staying in the driver’s seat. Start with something you know a bit about instead of being clueless on whatever about it. Why? So you can:
Spot when the AI’s off-base and fix it.
Add your own lawyerly spin—AI misses nuance.
Pick the best of its ideas and run with it.
Example: If you’re researching crypto regs, know enough to tell if a source is legit or if the AI’s hallucinating a dead link. I’ve logged over 1,000 hours with these models and taught hundreds of lawyers—this baseline keeps you safe and sane.
Vibe Working: Research Like a Rockstar
Now, the fun part: "vibe working." It’s you and AI riffing like creative buddies—natural, iterative, and way better than typing stiff prompts like a robot. The goal? Slash time on tasks you’d love to do but can’t—like a curated market brief—without risking your legal work product. Here’s how it works, step by step.
The Problem: Say you’re a litigator tracking blockchain and crypto disputes. Your inbox is a mess—Google alerts, firm ticklers, client updates—all generic, no insight. No one’s got time to sift through it, and you’re stuck.
The Vibe Fix: Use a reasoning model to build a weekly brief in 20 minutes that’d take hours otherwise. Here’s what I told Grok (in DeepSearch mode - it also has a DeeperSearch mode) on my phone, talking naturally:
Wait did you get that, I used my phone by just talking my prompt. I talk normally and stream of consciousness.
“It’s March 26, 2025. I need a weekly briefing for a law firm partner who litigates blockchain, crypto, and crypto-asset disputes. Focus on U.S. regulatory updates, industry news, and developments he can share with clients. Keep it current—last week only.”
Grok thought it over, asked if I wanted federal or state focus (I said both), then scanned 100+ webpages in minutes. It spat out a report with links—most real, some duds (e.g., SEC homepage instead of a specific release). I switched to Gemini’s DeepResearch, reused the prompt, and got better links, though some were older than a week. With my baseline knowledge (I dabble in crypto), I weeded out the junk, polished it, and had a tight brief in 20 minutes—better than any news feed.
Check out the briefing HERE:
Why It Works: Reasoning models dig deeper than chat models, pulling real-time info chat can’t touch. Talking to them feels human—no prompt engineering BS. And since it’s not legal work product, mistakes don’t tank you—they just teach you.
Your Playbook: Start Today
Here’s how to vibe your way ahead, whether you’re a novice or pro:
Grab the Apps: Download ChatGPT (OpenAI), Gemini (Google), and Grok (xAI) on your phone and each allows you to use your mic to talk to them. They’re free to start (some features vary by region—check restrictions).
Chat First: Play with ChatGPT. Ask it to summarize an artice or brainstorm client pitch ideas. Get comfy talking to it—voice mode’s a game-changer.
Reason Next: Try Grok or Gemini’s DeepResearch. Throw it a research task like my crypto brief. Check the links, tweak the prompt, repeat till it sings.
Vibe It Up: Talk naturally—say what’s on your mind, what you know, what you need. Example: “I’m no expert, but I need a rundown on AI patent trends this month—go.” The AI adjusts to you.
Share the Wins: Turn that brief into a client email or team update. Show off without breaking a sweat.
The Payoff:
Two years ago, this took me hours—now it’s 20 minutes, and the quality’s sharper. Novices, this builds your AI chops safely. Intermediates, it scales your game. Pros, it’s a new edge. No one’s helping lawyers with this stuff—firms are too busy, in-house teams are swamped. Vibe working fills the gap. So download those apps, talk to your AI crew, and leapfrog the pack. The future’s here—vibe your way to the top.

AI Updates: What’s New and What It Means
AI models keep evolving, and the past few weeks have brought some practical upgrades worth knowing about. Here’s a rundown of the latest features from OpenAI, Google, and Anthropic—explained simply for busy legal professionals.
OpenAI’s 4o Image Generation
What’s New: As of March 25, 2025, OpenAI’s 4o model ( powering ChatGPT and Sora) can now create images from text prompts. Type something like “a courtroom sketch of a judge,” and it generates a picture. It’s open to all users, free or paid.
Google’s Gemini 2.0 Flash Experimental
What’s New: About a week ago, this fast, lightweight model added native image generation. You can test it in Google AI Studio, and it now also pulls insights from YouTube video links you share.
Google’s Gemini 2.5 Pro Experimental
What’s New: Launched March 25, 2025, this is Google’s smartest model yet. It’s great at reasoning (thinking through problems step-by-step), coding, and handling text, images, audio, and video together. It’s available in Google AI Studio and for Gemini Advanced subscribers.
Google’s Gemini Canvas and Audio Overview
What’s New: Recently added to the Gemini family, Canvas likely lets you create or edit content (think documents or visuals), while Audio Overview summarizes or analyzes sound files—like a podcast or recorded statement.
Anthropic’s Claude Web Search
What’s New: Claude, from Anthropic, now searches the web to answer questions with up-to-date info. This rolled out in the last week or so.
Why It Matters for You: These tools are getting smarter and more practical, offering ways to cut busywork and boost insight. For legal pros, they’re best as assistants—think research, drafting, or visualization—not decision-makers. Responsibly, that means testing outputs, knowing their limits, and keeping human judgment in charge. Stay curious, but stay sharp.
That’s it. I promise to publish more frequently as many of you have reached out to ask.
To read previous editions, click here.
Was this newsletter useful? Help me to improve!With your feedback, I can improve the letter. Click on a link to vote: |
Who is the author, Josh Kubicki?
I am a lawyer, entrepreneur, and teacher. Not a theorist, I am an applied researcher and former Chief Strategy Officer, recognized by Fast Company and Bloomberg Law for my work. Through this newsletter, I offer you pragmatic insights into leveraging AI to inform and improve your daily life in legal services.
DISCLAIMER: None of this is legal advice. This newsletter is strictly educational and is not legal advice or a solicitation to buy or sell any assets or to make any legal decisions. Please /be careful and do your own research.