Over the weekend, a post on X by Zach Shapiro, a former big law lawyer who now runs a boutique practice, blew up. It pulled in millions of views and hundreds of bookmarks.
At a surface level, his message is straightforward. If you treat Claude like a real work surface, then wrap repeatable workflows around it, a small team can produce an outsized amount of legal work. You are encoding judgment.
I’ve already seen a few sharp takes. Artificial Lawyer calls this the legal world waking up to new workflow primitives—skills, agent mode, the rest. And over at LawWhatsNext, they call it “Claude maxxing”: domain experts turning AI into a personal OS for getting things done.
It got me thinking. Why did this gather so much attention?
The clearest answer I have landed on is a Moneyball analogy. Firms will not win by having better lawyering alone. Many already do. Firms will win by running a better system, measuring the right things, and converting cheap execution into reliable outcomes. Being operationally excellent.
Why this feels like Moneyball
Moneyball, in plain terms, is the idea that a team can outcompete bigger budgets by measuring the right inputs and building a repeatable system. (If you haven't read the book or watched the movie, it's worth it).
That is what this moment feels like. Many firms already have excellent lawyers. That will remain table stakes.
The edge shifts to operational excellence. Firms win by building a system that produces consistent outcomes at speed, with clear visibility into work, quality, and economics. They win via ops metrics, marketing ROI, and scalable delivery, not by claiming they “lawyer” better in the abstract.
In that spirit, I want to look at five facets of this shift. The cost of intelligence is going to zero. Managing AI agents as the next job. Practitioners will need craft knowledge to brief them. The biggest gains will go to people with deep domain judgment to encode. And teams with better AI configuration will move faster, at a cost structure that was previously unavailable.
1. The cost of intelligence is going to zero
For most of modern legal services, intelligence has been expensive because expert cognition is scarce.
You can scale scarcity, but you pay for it. You hire. You train. You manage. You absorb overhead. You accept latency.
AI has been impacting the cost curve. As models improve and workflows become agentic, the marginal cost of competent first-pass work trends toward zero.
Drafts, summaries, issue lists, clause comparisons, research plans, and first cut redlines. You can generate more versions at near-zero incremental cost. This does not mean the work is perfect. It means the firm can iterate faster than it could before.
What matters is where this is heading. When the slope points down, the firm that learns to convert cheap execution into real outcomes wins.
2. The next human job is managing AI agents
When execution gets cheap, people stop getting paid for doing the work and start getting paid for directing the work.
That is the shift sitting under the “Claude native law firm” narrative.
The highest leverage professionals will behave like managers of agents.
They will delegate, steer, review, and decide.
They will spend less time producing the first artifact and more time shaping the outcome.
This is a meaningful change in daily work.
Many knowledge workers have never had to write precise briefs. In many situations, we rely on other humans to fill gaps, to read between lines, and to ask clarifying questions.
Agents do what you ask, at speed, and eventually autonomously.
So the unit of leverage becomes the brief and instructions.
3. Practitioners need craft knowledge to brief agents
Here is the uncomfortable question that keeps coming up: can you manage what you do not understand?
In law firms, much of the management sits one layer above the work. That can work when the work moves slowly, and juniors learn through apprenticeship.
Agentic work breaks that comfort. To brief well, you need craft knowledge.
Not because you must do every task yourself. Because you must set constraints that reflect real risk, and you must recognize when output fails.
Judgment lives in the choices. Understanding risk tolerances, commercial posture, and many other variables.
So leaders who want the full value of agents must stay close enough to the craft to steer and evaluate.
4. The people who win are the ones with judgment to encode
This is why the Zach Shapiro story resonates with practitioners.
The advantage is not “Claude can do the work.”
The advantage is that a practitioner with deep domain judgment can encode that judgment into reusable instructions. That is what turns a model into a system.
The best outputs come from people who already know what good looks like.
They know the patterns. They know the edge cases. They know what must not happen.
They can translate that knowledge into constraints, checklists, and review rules.
This creates a widening gap.
A junior can use AI to move faster. A senior can use AI to multiply judgment.
The firm that helps seniors encode judgment and helps everyone reuse it compounds learning.
5. Configuration becomes a competitive advantage
Access to AI will not be the advantage - it's already commoditized.
The advantage will be configuration, solutioning, and setup.
The team with a better AI system can move faster and operate at a cost structure that was not previously available to it.
This system includes how you brief, what context you provide, how you review, when to escalate, and how you store what works.
This is the core of the cost curve.
Intelligence and execution trend toward zero marginal cost. The firm that captures that drop can expand output without expanding headcount at the same rate.
That changes pricing power. It changes responsiveness. It changes client expectations. It also changes how competition feels and who you compete with.
A small team can compete with a larger team because it can iterate faster. Not because it becomes smarter than the market. Because it becomes faster at turning judgment into outcomes.
The hidden cost is coordination
There is a predictable second-order effect.
Agentic tools increase output volume. But volume creates a new bottleneck.
Review becomes a constraint. Coordination becomes a constraint. Tracking becomes a constraint.
If you double or triple the amount of work product that can be generated, you also increase the amount of work that must be triaged, assigned, checked, and closed.
Software teams have already seen this.
Agentic coding increases throughput and the volume of work that requires management. More changes. More branches. More reviews. More merge conflicts. More decisions.
Legal and other knowledge-based work will follow the same pattern.
That is why grounded workflow matters more as intelligence gets automated.
When execution accelerates, you need a system that keeps work real.
You need a place where work can be tracked accurately, owned clearly, and audited cleanly.
As agents increase the pace of drafting, redlining, research, and client comms, the firm needs stronger scaffolding around the work. A task and matter system of record prevents the output surge from turning into noise. It keeps accountability intact. It keeps status honest. It keeps review loops visible.
AI can accelerate the work. A grounded system keeps the work coherent.
What I take from the Zach Shapiro moment
I come away optimistic.
It highlights the importance of leverage and how that can be achieved through [autonomous] systems.
It makes “small team, big output” feasible for practitioners who bring judgment and who put in the work to encode it.
It also makes the business of being a lawyer look different.
The firms that win the next decade will continue to have great attorneys. But they will also need to have an operating system.
They will treat SOP as a core skill. They will keep craft knowledge close to leadership. They will encode judgment. They will invest in workflow systems that can handle the new volume.
Moneyball did not reward the teams with the most tradition. It rewarded the teams that learned how to measure, decide, and execute differently.
Law is entering that kind of era.