Oz Benamram
AI and the Mid-Market Law Firm
Notes from a Food for Thought dinner hosted by Oz Benamram at BLACKBARN, in partnership with Lucio
For the past few years, the legal AI conversation has been dominated by the largest firms in the market.
That is understandable. The AmLaw 50 have the budgets, innovation teams, and scale to experiment early. They also have the client pressure to show that they are taking AI seriously.
But at a recent Food for Thought dinner hosted by Oz Benamram at BLACKBARN, in partnership with Lucio, the conversation turned to a different and arguably more interesting question:
What does AI mean for mid-market law firms?
Around the table were leaders from mid-sized firms, each wrestling with the same set of issues: which tools to trust, how to drive adoption, how to speak to clients, and how AI might reshape the economics of legal services.
The discussion made one thing clear. For mid-market firms, AI may be one of the most important competitive opportunities of the next decade.
Mid-Market Firms May Have an AI Advantage
Large firms have more resources, but they also have more bureaucracy.
Mid-market firms can often move faster. They can identify specific use cases, pilot tools quickly, make decisions without layers of committee approval, and push adoption across the firm with a more personal approach.
That matters because AI implementation is an organizational change project. The firms that succeed will be the ones that embed AI into everyday legal work.
For mid-market firms, this creates a meaningful opening. Tasks that once required large teams, especially document review, due diligence, research, first-pass drafting, and knowledge retrieval, can now be performed with far greater leverage. Work that historically favored firms with 20 junior associates available at short notice may now be within reach for smaller and more focused teams.
AI increases the importance of supervision, strategy, risk assessment, and client advice. The scale advantage of BigLaw is no longer as absolute as it once was.
General AI Tools Are Powerful, But Legal Work Needs More
One major theme from the dinner was the distinction between general-purpose AI tools and legal-specific platforms.
Everyone in the market understands that models like Claude, ChatGPT, and Gemini are extremely capable. They are improving rapidly and can produce impressive results across many tasks.
Beyond raw model capability, law firms require security, support, predictable pricing, and legal-specific workflows.
They need security. They need privilege protection. They need predictable pricing. They need workflows that reflect how lawyers actually work. They need auditability, citations, document handling, and support from people who understand legal urgency.
One example discussed at the dinner captured the pricing problem sharply: a $200/month Claude subscription that still generated $3,320 in token spend in a single week.
For law firms, that kind of unpredictability is difficult to manage. Legal work often involves long documents, heavy context, multiple iterations, and time-sensitive matters. Usage can escalate quickly.
General models are incredibly capable, but law firms need infrastructure, support, and cost controls around those models. A firm does not want to discover a problem with its AI tool during a live matter and have no one to call.
Model-Agnostic Is Becoming the Sensible Default
Another point of agreement was that no single model will permanently dominate every legal task.
One model may be better at long-context analysis. Another may be better at drafting. Another may perform better on speed, cost, or structured extraction.
That means firms should be cautious about locking their knowledge, workflows, and data into one model ecosystem.
The better architecture is model-agnostic: keep the firm’s data, documents, templates, and institutional knowledge separate from the underlying models used to analyze them.
This gives firms flexibility. As models evolve, the firm can use the best model for the task without having to rebuild its entire AI strategy.
For law firms, the long-term asset is not a single model. It is the firm’s own knowledge base, workflows, drafting standards, and accumulated expertise.
Adoption Is Still the Hard Part
The hardest part of AI is not access. It is adoption.
Most lawyers are busy. They bill by the hour. They are trained to be cautious. They do not want to spend time experimenting with tools unless the benefit is immediate and obvious.
That is why broad, open-ended experimentation rarely works.
The firms seeing real adoption are taking a more structured approach. They identify 10 to 15 specific use cases. They train lawyers one-on-one. They focus on workflows that already exist. They use peer success stories to build confidence.
This is very different from simply announcing that the firm now has an AI tool.
At the dinner, Falcon Rappaport’s reported 80% daily usage stood out because it reflected exactly this kind of disciplined implementation. Adoption came from clarity, training, and relevance to actual work.
The lesson is simple: lawyers adopt legal AI when it makes their work easier, faster, more consistent, or more defensible.
Client Conversations Are Changing
Client questions about AI are also evolving.
Initially, many conversations focused on cost savings. Clients wanted to know whether AI would make legal work cheaper.
That question still matters, but it is no longer the only one.
Increasingly, clients want to know whether AI improves the quality of legal work. Does it reduce missed issues? Does it help lawyers analyze more documents? Does it improve consistency across work product? Does it help firms identify risks earlier?
Responsible use is becoming part of procurement.
Clients may not always demand that firms use AI. But they will increasingly expect firms to have a coherent answer about how AI is governed, where it is used, how outputs are checked, and how confidentiality is protected.
For law firms, this creates both pressure and opportunity.
A weak AI story may soon be a disadvantage. A thoughtful AI story may become a differentiator.
The Demand Question: Where Does the Work Go?
One of the most important parts of the conversation was about demand.
AI will not affect all legal work in the same way. Its impact can be grouped into five broad categories:
Business teams self-serving and never reaching out to legal.
In-house legal teams expanding their own capacity.
In-house teams bringing work back from external firms.
Law firms doing existing work faster and cheaper.
Law firms offering new capabilities that were previously not practical.
The first three are the risk zone for law firms. They represent work that may never reach outside counsel.
The last two are the opportunity.
If a firm can use AI to deliver faster, improve accuracy, reduce cost, and offer new forms of analysis, it can strengthen its client relationships rather than weaken them.
The challenge is that this requires firms to rethink how they package, price, and communicate their work.
The Billing Model Tension Is Real
AI creates an obvious tension with hourly billing.
If a task used to take 10 hours and now takes two, the traditional billing model can punish the firm for being efficient.
That tension is not fully resolved.
Some work will still justify hourly billing, especially where complexity, uncertainty, and judgment dominate. But other categories of work may move toward subscription, fixed-fee, success-based, or value-based models.
The billable hour will not disappear overnight. However, firms need to be more intentional about where they create value. If AI reduces the time taken to produce a first draft or review a large set of documents, the firm needs a pricing model that reflects the value delivered, not merely the hours spent.
Mid-market firms may be particularly well-positioned here because they can experiment with alternative pricing more quickly than larger firms.
Build vs. Buy: Rent, Do Not Build
The room also discussed whether firms should build their own AI tools.
The practical consensus was: rent, do not build.
For most firms, building internally creates a maintenance burden. It often depends on one or two internal champions. It requires ongoing model updates, security reviews, workflow design, user support, and product iteration.
That is not usually the core business of a law firm.
There may be exceptions. If a firm has a truly unique workflow or proprietary data advantage that represents a competitive moat, building may make sense.
But for most use cases, vendor relationships are the more sensible path. The key is choosing vendors who are responsive, legally informed, and willing to work closely with the firm after the contract is signed.
In the mid-market, that support layer matters enormously.
The Real Opportunity: AI That Reflects the Firm
Perhaps the most interesting idea from the dinner was that legal AI should not feel generic.
The goal is not simply to give every lawyer access to a chatbot. The goal is to create systems that reflect the firm itself: its templates, clause positions, drafting style, prior work product, playbooks, risk preferences, and institutional knowledge.
In other words, the opportunity is the creation of a firm-specific intelligence layer.
That is especially important for mid-market firms, where consistency, leverage, and institutional knowledge can become major competitive advantages.
The firms that succeed with AI will not be the ones that ask, “Which model should we use?”
They will be the ones that ask, “How do we make our firm’s knowledge more accessible, repeatable, and valuable across every matter?”
Where This Leaves Mid-Market Firms
The dinner made clear that mid-market firms are not behind in the AI conversation.
In some ways, they may be better positioned than larger firms.
They can move faster. They can focus on practical use cases. They can train lawyers more personally. They can experiment with pricing. They can build deeper vendor relationships. And they can use AI to compete for work that once required far larger teams.
But the opportunity will not materialize automatically.
But for mid-sized firms to seize the opportunity there needs to be strategy, governance, training, use-case discipline, and a clear view of how AI changes the firm’s value proposition.
AI is forcing every mid-sized firm to confront a sharper question:But it may force every mid-market firm to answer a sharper question:
Are we using AI merely to defend our existing work, or are we using it to become a more competitive version of ourselves?
