The Edge and the Market

Nobody in the admissions industry will say this plainly, so I will.

AI tools are changing legal work, and that change has opened a window the right kind of law graduate can use. The window is open right now, in 2026. It won’t stay open forever.

The people who understand this and move on it will look back the way people look back at 2009 real estate prices. The rest will wonder what happened.

Here’s what the window looks like, and where to plant yourself inside it.

The window opened because of a gap. AI can do a lot, but the profession doesn’t yet know how to use it safely.

As of mid-2026, AI can cut a good chunk of legal research time. Not all of it, and not the judgment at the end. But the mechanical work, pulling cases, summarizing a field, drafting a first pass, doing first-round contract review, the kind of work that used to eat 6 to 8 hours of a first-year associate’s week, can now be cut by roughly 50 to 60 percent.¹ That’s a real change. Firms that ignore it will lose to firms that use it, and everyone in the industry knows this.

They’re slower to admit the next part. The tools are confidently, dangerously wrong, in ways that take a trained lawyer to catch. Not often. But the failures show up in exactly the kind of legal question that matters most. A case distinguished without being overruled. A statutory carve-out buried in a cross-reference. A factual detail the AI got backwards but stated with total confidence.

A first-year who treats AI output as a finished product is signing their name to the model’s mistakes. A partner who can’t check the output has no way to know that’s happening.

So the profession is in an odd spot. The tools are powerful enough to matter, but knowing how to use them safely is still rare enough to be valuable. That’s the window.

The skill that fills it is verification, and no, that’s not the same as prompting. You can learn prompting in a weekend, and everybody has it now. Verification is the legal judgment to read AI output critically, to know which flags to run down, to understand what the model usually misses and why. That skill builds on itself. A lawyer who starts building it in 2026 and 2027 will have 2 or 3 years of real expertise by the time the market figures out it wants to pay for it.

Maya is a composite, but her clinic story is built from a real pattern. In her second year, she used an AI research assistant on a housing-discrimination brief for a clinic client. The tool did the first case-law sweep in about 40 minutes. She had set aside 2 hours, so she spent the extra time on something she wouldn’t have done otherwise. She ran a manual check of every cited case’s later history, pulled the 4 most important holdings, and read them in full.

She found 2 real problems the AI had missed.² The first was a case the tool cited for a point it didn’t actually support. A later ruling had narrowed the holding, in exactly the fact pattern her client’s case resembled. The AI cited the original case, not the one that narrowed it. The second was a regulatory interpretation her client had relied on. It had been quietly rescinded, with no public announcement. 1 of those misses would have muddied the brief. The other would have hurt her client’s position.

She wrote both catches down. When she talked to a hiring partner at a mid-size firm that summer, she didn’t open with “I used AI.” She walked him through the work. Here’s a research task. Here’s how I cut it by about 60 percent. And here are the 2 real issues the tool missed that I caught on review. 1 of them would have hurt the client. She had an offer by the end of that week.

The partner told her, bluntly, that most candidates who mentioned AI talked about how fast they could produce a first draft. None of them talked about what they’d found wrong with it.

That’s the edge. The window is real and open, and it’s still early enough that using the tools carefully sets you apart. Of course firms will get smarter and training programs will catch up. The committees writing AI use policies right now, mostly people who don’t fully understand the tools, will eventually employ people who do.³ When that happens, the edge closes, and what’s left is just the price of entry.

Build the skill now, before that happens, because it pays off over a whole career. In the stress test, Maya’s summer offer came in 2028. The people who waited until 2029 applied to firms that had already built their own AI training. Those firms knew what they needed and had started hiring for it. Same market, no edge.

Now the geography. Honestly, this is where most applicants get it exactly backwards.

Rankings are national. Hiring is local.

You can use a national ranking to compare schools. It tells you something about academic reputation and something about bar passage rates. It tells you almost nothing about whether you’ll get a job in the city you actually want to live in. Hiring pipelines, alumni networks, and the firms that recruit on campus are all regional, and the ties between a school and its local legal market took decades to build. A school ranked 38th nationally might be the top feeder to its city’s most important firms. A school ranked 22nd in a different city might place its graduates poorly there, because 3 higher-ranked schools in that same city have had 40 years of alumni at the partnership level.

The crisis didn’t land evenly. Part One showed you the institutional pressures, and they hit hardest in specific markets: regions with shrinking populations, flat or shrinking legal demand, and 30 to 50 years of graduates from the same schools filling the local bench and bar. The Northeast and parts of the Midwest fit this profile. Those markets were already tight before the forces in this book hit, and they’re tighter now.

The Sun Belt metros, the tech corridors, the energy hubs, the biotech clusters tell a different story. Austin, Phoenix, Miami, Denver, the Research Triangle. In these markets, legal demand has grown faster than local schools can supply graduates. The established firms there don’t have 4 decades of alumni from the same 3 schools blocking every path to partnership. The mid-market firms that hire most new lawyers are young enough that partner-level relationships are still there to build.⁴

Maya chose Austin over 2 higher-ranked Northeast schools. On paper the ranking gap looked bad, but the market gap mattered more. She was entering a market where firms genuinely needed junior talent, her school had a strong record placing graduates into the exact firms she wanted, and the city’s legal community was growing fast enough that showing up smart and ready to work was enough.

There’s a structural factor underneath the geography point, and most applicants and advisors miss it.

The talent pipeline in legal services has a gap at the mid-level, and you’re walking into it.

The great resignation years pulled mid-level associates out in numbers the profession is still absorbing. Attorneys 6 to 9 years out normally bridge between senior partners and junior talent. They build institutional knowledge and train the new associates. Between 2021 and 2023, they left in unusual numbers, and partners at large and mid-size firms retired earlier than expected. A law degree takes 3 years to earn. A lawyer with 5 years of real experience takes 8. You can’t make that happen faster. And AI, while it automates the most mechanical junior work, hasn’t produced a single experienced practitioner.⁵

So firms need junior talent to fill the roles the mid-level gap created. The strategic graduate who walks into a growing market in 2026, 2027, or 2028 walks into real leverage. This isn’t speculation about some future trend. In the stress test’s timeline, Maya had 3 offers by September of her 3L year. She negotiated salary, remote flexibility, hours, and practice areas, and she could do that because the firms actually needed her and she knew they did.

She knew because she’d done the work to understand the market she was entering, instead of defaulting to the national ranking and hoping for the best.

The edge and the geography come together in a specific way for the 2026 applicant. The window on AI expertise is open and narrowing. The markets where the leverage is real can be found. They’re not secret, they just take actual research instead of a ranking lookup. The talent gap is a real force, and it favors the prepared graduate in a growing market.

These advantages build on each other. Think about two associates. One spent 1L clinical time building real AI-verification judgment, entered a growing market with clear eyes about where the hiring leverage was, and walked into 3L recruiting with that offer story ready. The other chased a prestige ranking into a saturated market with no edge. They land in very different places.

The next chapter is about building the foundation that makes all of that possible, because the edge and the geography are only leverage for the person with a real score. Without that, they’re just interesting facts about someone else’s career.

Watch: what's happening in the legal job market

The Law School Job Market Just Cracked

This chapter is from Anyone Can Beat the Law School Collapse by Steve Schwartz, LSAT Coach and Founder of LSAT Unplugged. I've been coaching the LSAT since 2005.

Published July 28, 2026.