AI-native law firms can dramatically lower the cost of legal services, but they also highlight deeper challenges around demand, attorney identity, and how future lawyers learn to develop judgment
Key highlights:
-
-
-
How AI-native firms redefine the lawyer’s role — AI-native firms like Paralex are gravitating toward a “technician” archetype, which puts less emphasis on the trusted-advisor dynamic that has long defined the attorney-client relationship.
-
The profession may be heading toward a two-tier split — As AI-native firms grow and normalize this operating model for a new generation of attorneys, the legal profession may bifurcate into a smaller cohort of relationship-driven advisors who provide deep, context-rich counsel; and a larger pool of proficient, AI-assisted technicians working at high volume.
-
AI firms can highlight how future lawyers learn — AI-native law firms are elevating a long-standing mentorship gap that threatens to erode how the next generation of lawyers develop independent judgment; and addressing it will require both creative AI-assisted solutions and more deliberate frameworks for deciding which cognitive tasks should remain done by humans.
-
-
The opportunity of starting a native AI law firm to test an idea is intriguing to some lawyers, especially those with an entrepreneurial instinct and determination to see the idea through. When Stephen Candelmo founded Paralex.ai, he aspired to democratize legal services for small businesses and leveraged AI to do so. His 29 years of practicing law had shown him the inefficiency and costly downsides of the billable hour; and he hypothesized that if the workflow could be automated with an attorney in the loop and could charge one-tenth of what it normally cost, demand would follow.
The reality has been more complicated and more instructive for the future of legal practice, Candelmo explains, as he offered a candid accounting of what Paralex has learned in practice.
Building for underserved small business owners
Paralex was built around a tiered service model covering everything from verified legal Q&A to AI-assisted contract drafting. AI handles the intake and first drafts at every stage, and the attorney handles the judgment. The small business transactional law vertical was a deliberate bet because the practice area is most amenable to pattern recognition and workflow automation. In addition, small business represents one of the largest pools of underserved legal clients.
Candelmo has learned that affordability alone does not unlock demand. The long-cited statistic that “60% of small businesses never use a lawyer because of cost” overstates how much of that gap is price-driven. Indeed, a meaningful portion of business owners appear to not want legal counsel at any price. Free AI tools have compounded this learning because ChatGPT, Claude, and Gemini can produce a plausible contract or answer a legal question at zero cost. “People feel that maybe it’s just good enough,” Candelmo says.
How AI-native firms redefine concept of a lawyer
AI-native firms like Paralex are discovering they need to develop exclusively the “technician archetype” among its lawyers. The attorneys who thrive in Paralex’s workflow are those most comfortable operating at volume, untroubled by the absence of ongoing client relationships, and motivated by clean execution rather than the slower cultivation of client relationships. Candelmo describes them as comfortable with gig work because they want to be paid for what they produce rather than chasing invoices.
Young attorneys need to master the tools but not outsource their judgment to them. And they should seek out senior attorneys and cultivate human relationships that will make them more than a technician.
Candelmo shares that the trusted-advisor attorney who deeply knows a client’s business, anticipates problems before they arise, and provides counsel grounded in years of accumulated context is largely absent from the Paralex experience. He describes AI-native firms’ role as taking out the unnecessary back-and-forth that occurs in traditional law firms’ practices. At the same time, AI-native firms start out narrowly servicing a vertical by providing legal services that are optimizing for efficiency and relatively less complex.
The implication is significant for lawyers and their professional identity. As native AI firms grow and attract a generation of attorneys for whom this model is normal, the profession might be more likely to bifurcate between a smaller cohort of relationship-driven advisors on the one hand, and a larger pool of technically proficient, AI-assisted attorneys working at volume on another.
A generation of lawyers with no one to learn from
What Candelmo says he worries most about is who will teach the next generation of lawyers how to think. In AI-native environments, a junior attorney working at high throughput may review AI-generated output quickly, trust it, and move on. The output looks complete — but there is no obvious signal that something important was missing and no senior attorney to say why it matters.
Candelmo’s proposed solution is a second layer of AI tools, such as simulation tools, that can function like a senior lawyer. It reviews the initial output, flags gaps, and provides the kind of annotated feedback that would have come from a partner review in a traditional law firm.
His advice to young attorneys is to master the tools, but do not outsource your judgment to them. And they should seek out senior attorneys and cultivate human relationships that will make them more than a technician, Candelmo adds. “Ensure that your humanness, your human relationship skills make you stand apart.”
As native AI firms grow and attract a generation of attorneys for whom this model is normal, the profession might be more likely to bifurcate between a smaller cohort of relationship-driven advisors on the one hand, and a larger pool of technically proficient, AI-assisted attorneys working at volume on another.
In addition, David W. Simon, Partner at Foley and Gardner and adjunct professor at the teaches at the University of Wisconsin Law School, goes one step further and advocates for adding a conscious step before instinctively turning to AI tools. He suggests each lawyer first ask themselves, “What cognitive function is being delegated to GenAI at each step in the workflow?”
In the current state, the AI conversation within the legal ecosystem continues in a good-or-bad binary rather than simply asking when AI use is beneficial and when it is risky, which is increasingly what law students are asking for. For example, the University of UC Berkeley School of Law announced a policy that bans students from using AI for class assignments and during exams, although students can still use AI for research to identify sources.
The experiences of Candelmo and Paralex, alongside the broader debate playing out across the legal ecosystem, make it clear that the legal profession is being forced to make deliberate choices about what lawyers are for, which cognitive tasks should remain human, and how professional judgment is developed and passed on.
The law firms and legal institutions that build thoughtful frameworks for when and how AI should be used will create a profession that is both more efficient and more capable of producing the kinds of lawyers that clients and society will continue to need.
You can find more about the impact of AI in the legal industry here
