When Kevin Andrews was a child, he learned something troubling about his babysitter. She had stolen his mother's green card and was holding it for ransom. Years later, when his mother finally told him the full story, explaining why she had decided to pursue citizenship, young Kevin found himself captivated by questions that would shape his entire career: What is a green card? What is citizenship? What are visas?

"Fast forward a few decades after that," Andrews recalls, "my interest in international relations, politics and law school landed an opportunity to work specifically in the field of business immigration law. And that niche practice was something that I just became really interested and passionate about over time."

Today, after 15 years of practicing immigration law, Andrews has launched his own practice in Baltimore, focusing on employment-based immigration, including EB-1, NIW, O-1, and H-1B cases. But what distinguishes him from many of his peers is not just his legal expertise. It is his philosophical approach to artificial intelligence and his conviction that the legal profession is on the cusp of a transformation that will reward wisdom over raw intelligence.

Kevin Andrews on Why He Started His Own Immigration Law Firm

Andrews jokes that "entrepreneur is just a sexy French word that literally translates to undertaker." Like mortgage, he notes, we tend to invent elegant terms for difficult undertakings. But his decision to launch his own practice was driven by something more than ambition. It was sparked by his growing obsession with (and fear of) artificial intelligence and its potential to democratize capabilities that were once the exclusive domain of highly paid specialists.

"What excites me in the next 12 months is the ability to leverage AI to do things like code and to apply it to my work, to scale fast, getting weeks of work done in days.," Andrews explains. "I need to reassess the landscape the next 18 to 24 months. But I’m not really sure humans will derive basic meaning and purpose three to five years from now as innovation accelerates."

This is not the language of a technophobe clinging to traditional practice. Nor is it the breathless optimism of someone who believes AI will solve everything overnight. It is the measured assessment of someone who has spent considerable time thinking through both the promises and perils of intelligent systems.

How Government AI Is Reshaping Immigration Enforcement and Adjudication

When asked about AI's impact on immigration law by the end of 2026, Andrews pivots away from the private sector entirely. His concern lies elsewhere.

"My reaction to that question has more to do with the government’s increasing use of AI in investigations, enforcement and increasingly in adjudications," he says. "USCIS recently opened a vetting center in Atlanta that explicitly uses AI to adjudicate certain cases. These capabilities already exist and are operational now."

Andrews describes what he calls a "hockey stick sort of acceleration" in government AI capabilities, especially since early 2025. He points to the State Department's summer announcement on Twitter that they had revoked visas for individuals based on politically insensitive memes they had posted online.

"There was First Amendment outrage that happened," Andrews recalls. "But I'm like, what about the outrage over the government’s technical capabilities ? Did you guys know this is even possible, that they could hone in on that level?"

He describes ImmigrationOS, a system he has been following closely, as a "very user friendly dashboard where a human ICE agent can come in and say I want to create an agent that looks at all these social media to find a certain result." That agent can then hand off its findings to another AI agent that cross-references foreign nationals, which passes results to yet another agent checking IRS records, motor vehicle databases, and work authorization compliance in a sophisticated agentic workflow using Large Language Models like ChatGPT, Gemini, and Claude.

"What we see on the news in Ukraine and Iran and the Middle East, are tests for what they can and are doing domestically," Andrews observes. "It's just at first we get it as enforcement and investigation, but increasingly it creeps into adjudication." 

Kevin Andrews's Practical Strategy for Machine-Readable Case Filings

This awareness of government AI has already changed how Andrews practices law. He describes a pattern he has noticed among colleagues receiving RFEs claiming they failed to provide certain evidence.

"The RFE says a letter or something wasn’t in the initial filing and I can see that it was submitted on page 742," he says. "I was initially thinking this is a human training error. Then I started thinking it's actually a machine readability error."

His research into the DHS AI case system led him to restructure his cover sheets. "Back in the day we would say, you’ve got to put a red cover sheet on something to draw attention to an officer," Andrews explains. "Maybe in the near future we’ll need to put a QR code in to draw attention to the human officer. But for now it has to do with the font and the content of what goes on the exhibit cover sheet."

This practical adaptation exemplifies Andrews's broader philosophy: understand what the technology is doing, then adjust your practice accordingly.

The Paperclip Paradox and Why Wisdom Matters More Than Intelligence

Andrews's thinking about AI extends far beyond immigration law into deeper questions about the nature of intelligence and wisdom. He invokes the famous "paperclip paradox," a thought experiment about an AI instructed to maximize paperclip production.

"If you tell an AI to make paperclips and you don't give it a purpose, it'll convert the entire universe into paperclips," Andrews explains. "It'll just destroy everything because its only purpose given was to make paperclips. That is intelligence, the ability to do something that is completely devoid of any wisdom."

He notes that AI IQ was around 80 or 90 in 2024, surpassed average human IQ in 2025, and current high-end models operate around 140 to 160.

"So what happens when it's 200 and 2,000 and 2 million?" Andrews asks. "When intelligence is an abundant utility like Sam Altman is saying,  harnessing that intelligence with wisdom is what will matter."

This is where Andrews believes human practitioners retain irreplaceable value. "The wisdom part, at least this iteration of AI, doesn't have it. And it can only act when prompted either by words or triggers."

The Steak Metaphor: Why AI Needs Human Digestion

Andrews has a vivid way of describing what he sees as wrong with many current AI solutions for legal work. He calls it the "steak metaphor."

"Many AI solutions providers show us this delicious steak but they just shove it in your mouth all the way through your body without digesting it.  It doesn’t matter how good the food (output) is if I can’t process it. There's a lot of interfacing that I need to do with that steak before I can confirm whether it’s delicious ."

What Andrews wants instead is modular, iterative workflows with AI tools. For EB-1 case analysis, for instance, he envisions working through criteria one at a time, committing work on each criterion before moving to the next, building a context that persists across sessions.

"The point is to make this a more efficient workflow, not to make it an instant workflow," he says. "There's going to be compounding hallucination if I don't check it in an iterative and methodical way."

He adds: "Whoever figures out how to make that workflow where the interfacing feels almost as natural as working with a brilliant assistant without anthropomorphizing it will capture a lot of wealth in this market in the near-term."

Kevin Andrews's Advice for Immigration Lawyers: Start with Your Kitchen

Andrews's advice for fellow immigration attorneys is unexpectedly domestic. He recalls a moment last summer when it dawned on him that he could simply point his phone at his refrigerator and ask Gemini what he could make for dinner.

"The response that it gave me: show me your spice cabinet first," he recalls. "And then it said, do you want to make something simple, healthy, and gave me several options. A switch went off and I realized just how general purpose this technology is."

His point is that attorneys should become "AI generalists" by applying these tools to everyday life before bringing them into practice.

"My advice is to start thinking about applying it to other aspects of your life and to maintain a healthy sense of intellectual humility."

He extends this philosophy to his children, ages six and nine. "I get them to vibe code their own video games instead of just playing regular games," he says. "I also think an education focused on the  humanities and understanding human history is so much more important now than ever because that's how they will acquire the wisdom to deal with AI without any work experience."

In Andrews's view, a software engineer with great coding skills and lacking a sense of philosophy will not survive the new economy. "But the inverse will not be true," he argues. "A software engineer that reads Marcus Aurelius and the Bhagavad Gita but can’t code by hand  will probably thrive in the new economy."

The Twenty-Year Question: What Problems Have You Always Wanted to Fix?

For practitioners who have spent decades in immigration work, Andrews offers a reframe that leverages their experience rather than rendering it obsolete.

"If you've been a knowledge worker for the last 20 plus years, you’ve had a situation where you say, how come this doesn't automatically go over here? Why doesn't this do this? Why do I have to copy and paste this here and here in this system?" Andrews notes. "That used to be a complaint to a co-worker, maybe even a tech request, but now it can be a vibe code project to fix that thing in an afternoon."

He continues: "The people who have the workflow experience over the last 20 years have the best answers to the question, “What should I build?”"

How Kevin Andrews Achieved 1.2 Million LinkedIn Impressions with AI

Five months into running his own practice, Andrews identifies marketing as the area where he has seen the largest return on investment from AI. His LinkedIn impressions over the past year have reached approximately 1.2 million, representing a growth rate of more than 43,000 percent over the prior year.

""I've been doing consultations with people for two to three hours a day, five days a week for 15 years. That gives me a sense of what people are interested in and that’s what I write about."

His approach is not to ask AI for generic content. Instead, he has conversations with AI about topics he already understands deeply, using research tools to refine and verify his insights, then leveraging AI-generated images and infographics to draw attention to his content.

Looking Ahead: Kevin Andrews on the Next 18 to 24 Months

Andrews is careful about making predictions beyond a certain horizon. "I got to reassess because I'm not really sure after that," he admits when discussing anything beyond the next two years. But his near-term vision is clear.

"This mindset that you can fix problems that involve coding but doesn't necessarily involve having to know how to code anymore," he says. "Focus on workflow optimization and then start with the things that are the most annoying but easiest things to fix."

He notes that IBM recently tripled its hiring for entry-level workers with AI generalist skills. The prompt engineer role that commanded $300,000 to $400,000 salaries in 2020 and 2021 is now as basic as knowing how to use Microsoft Word was in 2015.

"Setting up a database called a RAG, a retrieval augmented generation pipeline, is something you can just sort of vibe talk to Claude about now," Andrews observes. "Two years ago you might have needed to understand things about machine learning to accomplish this. But then something called Model Context Protocol (MCP) made it super easy to do without the need to fully understand how."

For Kevin Andrews, the future of immigration law is not about replacing attorneys with algorithms. It is about attorneys who understand both the technology and the timeless need for human wisdom, who can work alongside AI systems that are increasingly sophisticated without losing sight of why that work matters in the first place.

After all, somewhere out there is another child hearing about green cards for the first time, asking the same questions Andrews asked decades ago. The technology to answer those questions will keep advancing. The wisdom to answer them well will require a human touch for now.