A Client Is Not a KPI: The Danger of Assembly-Line Artificial Intelligence in Law
Artificial intelligence should give lawyers more time to represent people — not give law firms a faster way to process them.
A national personal-injury firm recently issued a breathless announcement about its new proprietary artificial-intelligence platform. The promotional materials emphasized the enormous amount of money already invested, a commitment to spend more than a billion dollars over the next decade, thousands of active users, millions of AI interactions, and dramatic increases across certain “key performance indicators.”
It was undeniably impressive as a technology announcement.
But was it impressive as an announcement about client representation?
That is a different question and it is the one lawyers should be asking.
The Problem Is Not Artificial Intelligence
Let me be clear: I am not opposed to artificial intelligence. Law firms should use modern technology. AI can help organize medical records, identify missing documents, prepare chronologies, locate information, reduce clerical work, and give attorneys more time to think about the issues that actually require professional judgment.
Used responsibly, technology can make lawyers more careful, more responsive, and more available to their clients.
The concern is not that a law firm built its own AI platform. A secure, closed system may provide greater control and better protection of confidential information than a consumer product. Nor is there anything inherently wrong with investing heavily in technology.
The concern is how success is being defined.
When a law firm promotes its AI by emphasizing money spent, interactions completed, demands generated, discovery responses produced, and increases in staff efficiency, it is describing the productivity of a system. It is not necessarily describing the experience — or the outcome — of a client.
Volume is an operational metric. It is not a measure of justice.
More Output Does Not Necessarily Mean Better Representation
A billion-dollar technology commitment proves that a firm has a large technology budget. It does not prove that an injured client received better representation.
Millions of AI interactions prove that a system is being used. They do not establish that its conclusions are accurate.
More demand letters sent and more discovery responses filed demonstrate throughput. They do not tell us whether a client’s calls were returned, whether the client ever developed a meaningful relationship with an attorney, whether the lawyer understood the human consequences of the injury, or whether the case produced a stronger result.
A demand letter is not better simply because it was produced faster. A case has not necessarily been valued fairly because an algorithm compared it to thousands of other cases. A file moving quickly through a system is not the same thing as a person moving closer to justice.
If technology improves client satisfaction, strengthens recoveries, reduces mistakes, and gives attorneys more time to communicate with the people they represent, those are accomplishments worth celebrating. But those results require evidence. They cannot simply be inferred from how many documents a platform produces.
Clients Are Not Standardized Units
Standardization has a legitimate place in a law office. Deadlines should not be missed. Records should be requested promptly. Documents should be organized consistently. Routine administrative work should not consume time that could be spent counseling clients or developing cases.
But the value of a personal-injury case is not found on an assembly line.
Every client brings a different medical history, family situation, occupation, recovery, venue, tolerance for risk, and experience of loss. Two people with similar diagnoses may live with entirely different consequences. A historical database can identify patterns, but it cannot experience pain, understand fear, or recognize every way an injury has altered the course of a particular life.
Patterns should help a lawyer ask better questions. They should not become a substitute for asking questions.
There is also a deeper concern whenever an organization trains a proprietary system around its own historical practices. The system may learn the organization’s strengths, but it may also reproduce its assumptions, habits, valuation tendencies, and blind spots. Repetition does not automatically transform an institutional practice into the right practice.
Consistency can prevent careless errors. It can also standardize them.
“Proprietary” Does Not Mean “Proven”
An internally developed platform may be excellent. It may be secure, carefully tested, and rigorously supervised. But the word “proprietary” does not establish any of those things.
Public announcements filled with impressive usage figures still leave important questions unanswered:
What are the platform’s independently tested accuracy and error rates?
How are AI-generated medical summaries and factual statements verified against the source documents?
What actions can an AI agent take before an attorney reviews its work?
Are mandatory human-review checkpoints built into every consequential workflow?
How are mistakes identified, reported, and corrected?
What safeguards prevent information from one client’s matter from improperly affecting another?
Are clients informed when AI plays a meaningful role in drafting, analyzing, or communicating about their case?
Has anyone independently verified that the claimed efficiency gains actually produce better client outcomes?
The absence of those answers from a promotional launch does not prove the safeguards are missing. It does, however, mean that spending, scale, and self-reported productivity should not be mistaken for demonstrated reliability.
The Lawyer Remains Responsible
The Alabama State Bar addressed these issues directly in its 2026 guidance on artificial intelligence. It emphasized that lawyers remain fully responsible for work performed on behalf of a client, regardless of whether AI helped create it. The guidance warns that agentic AI (e.g. technology capable of planning and executing tasks with minimal human involvement) creates particular risks because an unsupervised mistake can produce real-world consequences.
The Bar recommends documented human-review checkpoints before an AI system takes action in a client matter, regular audits of activity logs, independent verification of AI-generated work, and written limits on the tasks a system may perform autonomously. Most importantly, it states that legal judgment, negotiation, strategy, and client advice must not be delegated to artificial intelligence.
Those obligations do not disappear because the technology was expensive. They do not become less important because the software was developed internally. And they certainly do not diminish because a system can perform a task in seconds.
The faster technology operates, the more important meaningful human supervision becomes.
Is the Technology Serving the Client or the Business Model?
There is nothing improper about a law firm earning a profit. Law firms are businesses, and responsible investment in people and technology costs money.
But firms should be candid about who benefits from efficiency.
If artificial intelligence allows a lawyer to spend less time sorting documents and more time talking with a client, the client benefits. If it helps identify an overlooked medical issue or missing source of insurance coverage, the client benefits. If it reduces delays while preserving careful attorney review, the client benefits.
If, however, the principal result is that each lawyer can carry a larger inventory of cases, send more automated documents, and move more files with fewer human touchpoints, the principal beneficiary may be the firm’s balance sheet.
Calling that “innovation” does not answer the question.
That is particularly true when a law firm describes itself not merely as a legal practice but as an emerging technology platform that intends to sell its infrastructure throughout the industry. Perhaps that business will be successful. But success as a software company and excellence as a law firm are not the same measurement.
Clients should not be asked to assume that what makes a firm more scalable necessarily makes their representation more personal, more careful, or more effective.
Technology Should Make Lawyers More Available
The future of law is not a contest between lawyers and machines. Responsible law firms will use advanced technology, and clients should expect them to do so.
The real question is what the technology allows the lawyer to become.
Does it make the lawyer more informed? More prepared? More accessible? Does it create time to listen, explain, investigate, and exercise judgment? Or does it simply allow the organization to place more human beings into a larger and faster processing system?
Artificial intelligence can assist a lawyer. It should never become a system for processing human beings.
Efficiency is a virtue only when it gives a lawyer more time to know the client. If it merely allows a firm to carry more cases, generate more documents, and move more files with fewer human touchpoints, the principal beneficiary may not be the injured person. It may simply be the firm’s balance sheet.
Technology should make lawyers more available — not make lawyers less necessary.
Nothing herein should be contrued as legal advice, as to create an attorney-client relationship. This is for general, informational purposes only.
No representation is made that the quality of legal services to be performed is greater than the quality of legal services performed by other lawyers.
