Visalaw Team
Artificial intelligence is changing the legal industry at a remarkable pace. New tools appear constantly, each promising to save time, improve accuracy, automate workflows, or transform the way lawyers practice. For immigration lawyers, where firms manage high volumes of client data, government forms, deadlines, evidence, and case strategy, the promise of AI is especially compelling.
But as Colin Levy emphasized during a recent episode of Immigration Legal Tech Conversations, successful AI adoption is not simply about finding the newest or most powerful tool. It is about understanding the problem, preparing the people who will use the technology, evaluating vendors carefully, and making sure that accountability exists before, during, and after implementation.
As Levy put it, legal technology is still “an ecosystem of people, process, and tools together.” That framework is even more important in the age of AI.
One of the most important themes from the conversation was that law firms should resist the temptation to begin with the technology. The better starting point is the problem.
Levy warned against “a solution in search of a problem,” a common pattern in legal tech adoption. A firm sees an impressive demo, hears that a product is “AI-powered,” and assumes it must be useful. But if the firm has not clearly identified what it is trying to improve, the technology may create more frustration than value.
“You can’t solve a problem with a tool that isn’t designed to solve that problem,” Levy explained. “And you’re not going to know if you have a tool designed to solve that problem if you don’t know what the problem is.”
For immigration law firms, this is especially relevant. A firm may want to improve client intake, document collection, petition drafting, case tracking, deadline management, or internal knowledge sharing. Each of those problems requires different workflows, different data, different users, and different risk controls. A tool that looks impressive in a demo may not fit the firm’s actual needs.
Before evaluating an AI product, firms should define:
What problem are we trying to solve?
Who experiences that problem?
What does the current process look like?
Is this truly a technology problem, or is it a process or training problem?
What would success look like after implementation?
That last question is critical. Without a clear definition of success, firms may confuse activity with progress.
One of the strongest points in the discussion came from Levy’s writing on legal tech hype: “The gap between ‘we are implementing this technology’ and ‘we have changed how we actually work’ is where most legal tech initiatives go to die. Implementation is not adoption. Adoption is not transformation.”
That distinction matters.
A firm can sign a contract, migrate data, train staff, and technically “go live” with a new tool. But that does not mean the tool has been adopted. Adoption happens when people actually use the technology in their daily work. Transformation happens only when the technology improves the way the firm operates.
Many legal tech projects fail because leaders treat implementation as the finish line. In reality, implementation is the beginning of the hard part.
Lawyers, paralegals, legal assistants, and administrative staff need to understand not only how the tool works, but why it matters. They need to see how it fits into the firm’s workflow. They need to trust that it solves a real problem. And they need to have a voice in the evaluation process.
As Levy noted, when users are excluded, they may quietly avoid the new system or actively resist it. “You’re creating a new problem in addition to the existing problem,” he said.
Greg Siskind shared a practical example from the immigration law context: case management systems. For many immigration firms, case management software is not just one tool in the stack. It is the central system where client data is stored, immigration forms are generated, deadlines are tracked, and matters are managed.
Changing that system can affect nearly every person in the firm.
Siskind explained that after past experiences with technology transitions, his firm took a more deliberate approach when selecting a new case management platform. The firm created a committee that included administrative staff, paralegals, and attorneys. The group had time to evaluate products, attend events, and help shape how the eventual system would be customized.
That broader participation made a difference. The software worked better because the firm took the time to configure it around real workflows. Just as importantly, the process created buy-in across the organization.
This is a lesson every law firm can apply to AI adoption. The people who will use the tool should help evaluate it. Senior partners, firm owners, or technology committees may control the budget, but they should not be the only voices in the room.
A polished demo may appeal to leadership. But paralegals, associates, and support staff often know where the real workflow friction lives.
AI vendors often market their products with bold claims: faster drafting, better research, automated review, instant answers, intelligent workflows. Some of those claims may be legitimate. Others may be exaggerated. The only way to tell the difference is to ask specific, practical questions.
Levy described these as the “relentlessly boring” questions. They may not be exciting, but they are where real value and risk management begin.
For AI tools, those questions include:
What exactly does this tool do?
What does it not do?
How accurate is it in the specific use case we care about?
How much human review is required?
Where is client data stored?
Is our data used to train the model?
How long is data retained?
Who is responsible when the tool produces an error?
Can we export our data if we leave?
What support will we receive during implementation and termination?
These questions are especially important for lawyers because AI adoption is not just an operational decision. It is also an ethical and professional responsibility issue. Lawyers must protect client confidentiality, supervise technology use, verify outputs, and maintain independent professional judgment.
As Levy noted, “Exciting and valuable are two different things.” The information firms get from asking boring questions may not be flashy, but it is essential.
A vendor demo can be useful, but firms should not rely on demos alone. Demos are usually designed to show the product under ideal conditions. They may not reveal how the tool performs with the messy, inconsistent, high-stakes materials that lawyers handle every day.
Levy recommended that firms test tools against real or realistic scenarios tied to the specific problem they are trying to solve. That might involve a sandbox, a proof of concept, or a guided session where actual users test the tool with sample documents, workflows, or fact patterns.
For an immigration firm, that could mean testing whether a tool can help summarize a client’s immigration history, organize evidence for a petition, identify missing information in an intake, or draft a first-pass case memo. The point is not simply whether the tool can produce something impressive. The question is whether it produces something useful, reliable, reviewable, and aligned with the firm’s actual work.
A strong proof of concept should include the people who will use the tool every day. Their feedback should shape the final decision.
One of the most practical parts of the conversation focused on what happens when a firm needs to leave a vendor.
This issue is not unique to AI, but AI makes it even more important. Law firms may store sensitive client data, matter history, documents, work product, and internal knowledge inside vendor systems. If the relationship ends, the firm needs a clear path to retrieve that data.
Levy advised firms to ask about termination rights, data extraction, fees, timelines, vendor support, and retention policies before signing the contract. “You want to have a clean exit when that happens,” he said, “because you don’t want to be doing things by the seat of your pants.”
Siskind added that immigration firms have seen this problem before with case management vendors. Some firms have faced high costs or significant obstacles when trying to move their data to a new system. In some cases, vendors have been accused of effectively holding data hostage.
That is why data portability should be part of vendor due diligence from the beginning. Firms should know:
How do we get our data out?
In what format?
How long will it take?
What will it cost?
Will the vendor assist with migration?
How long will the vendor retain our data after termination?
Will the vendor delete the data upon request?
A vendor’s exit policies can reveal a great deal about its approach to customer relationships. As Siskind observed, making it easy for clients to leave can actually make them more comfortable choosing the vendor in the first place.
The conversation also turned to the next generation of lawyers and the role of judgment in an AI-enabled profession.
AI tools can generate text, summarize documents, identify patterns, and assist with research or drafting. But they do not think like lawyers. They can produce outputs that sound confident and plausible while still being incomplete, misleading, or wrong.
Levy cautioned that many AI tools are designed to be “people pleasing.” They may give users what they ask for, but not necessarily what they need. That is where legal judgment becomes indispensable.
Lawyers must understand the subject matter deeply enough to evaluate AI outputs. They must know when an answer is incomplete, when a legal issue has been missed, when facts require further investigation, and when a generated draft does not reflect the client’s actual needs.
This has major implications for training. If firms use AI to reduce the amount of work junior lawyers perform, they must still ensure that young lawyers develop the experience and judgment needed to become senior lawyers. AI can support training, but it cannot replace the process of learning how to think like a lawyer.
The profession should not assume that younger lawyers are automatically more prepared for AI adoption simply because they grew up with technology. Levy noted that the generational divide is often overstated. Some younger lawyers are eager adopters, while others are ambivalent. Some older lawyers are cautious, while others are leading innovation.
The real divide is not age. It is mindset, training, and intentionality.
The central takeaway from the conversation is clear: AI adoption succeeds only when people, process, and vendor accountability are aligned.
That means law firms should not treat AI as a magic layer that can be placed on top of broken workflows. If a process is unclear, AI may make the confusion more visible. If a team lacks buy-in, AI may increase resistance. If a vendor relationship lacks accountability, AI may introduce new risks around data, ethics, and dependency.
For immigration lawyers, the opportunity is significant. AI can help firms work more efficiently, manage information more intelligently, and improve the delivery of legal services. But the firms that benefit most will not necessarily be the ones that adopt the fastest. They will be the ones that adopt with discipline.
They will define the problem first.
They will include the right people in the process.
They will test tools against real workflows.
They will ask specific questions about accuracy, confidentiality, data use, and exit rights.
They will preserve lawyer judgment.
And they will remember that legal technology is never just about technology. It is about how people use tools to deliver better legal work.
That is where the real transformation begins.
View all blogs
Subscribe below to receive tips for using AI in immigration law, product updates, and relevant news in immigration.
Memphis, TN
.png)