The Impact of Legal AI: A Tale of Four Markets
The Pace of Change and the Rise of FOMO
The world is changing faster than most of us have become accustomed to, creating opportunities for some and significant FOMO for others. We are seeing this dynamic play out every single day in the legal industry. The narrative is everywhere:
"You have to adopt AI, or you'll be left behind."
"The future of legal practice is lawyers plus AI."
We get it. We understand why there is this sense of urgency around change.
This is an emerging technology that nobody really knows its limitations, boundaries, or full capabilities yet.
Take a breath. If we take a pause — look out the window at the flowers, the grass, the clouds in the sky — and if we talk to the actual end users of legal services, a different picture emerges. When we speak with the recipients and beneficiaries of legal services, we see something more nuanced.
The Distinct Markets
Focusing on transactional law, we are seeing a separation of the market between the top and the bottom.
The Low End: Commoditized Legal Services Turn to Self-Served
At the lower end of the market, businesses and ordinary consumers are making do with ChatGPT, Claude, and Gemini to satisfy their legal needs. These consumers of legal services are getting help understanding what they can and cannot do through general purpose chatbots. A handful of startups offering legal AI tools for consumers appear to lag in traction, possibly because they are competing indirectly against the general purpose AI assistants.
While these consumers may not always know the right questions to ask, the general purpose models provide “good enough” responses and directional guidance. As a result, lawyers as providers of legal services at this end of the market have seen (or will likely see much more of) their business decline.
The High End: Bet-the-Farm Work Face Steeper Competition
At the higher end of the market — where deals are worth millions and billions, the "bet the farm" work — people are still going to top-tier law firms. This segment is relationship-based, and quality remains paramount.
Two megatrends are emerging:
On the supply side, top tier firms are increasingly making use of AI in their workflows.
On the demand side, more in-house teams are performing the work internally with the assistance of AI.
On a superficial level, many of these AI tools perform adequately. But, when you dig deeper, the generative AI tools don't quite deliver quality and reliable work. Perhaps there is insufficient training data. Perhaps the retrieval augmented generation cannot pull back the right data. Perhaps they haven't built workflows for complex scenarios. Whatever the reason, these tools still require human review and corrections for complex work.
We know a percentage of the complex work are being done by AI without human review (see the news of Deloitte in Australia having to issue a refund for AI slop: https://fortune.com/2025/10/07/deloitte-ai-australia-government-report-hallucinations-technology-290000-refund/). We don't know the exact percentages, but we know a non-zero proportion of work is going straight from a chatbot to a client’s inbox. If lucky, there may be no consequences.
What All Clients Actually Want
We always return to the fundamental question: What do the customers paying for legal services actually want?
The buyers of legal services want reliable and correct legal advice. And with the raise of AI, these buyer of legal services want costs lowered and the response time shortened.
Fast, good, and cheap; can clients have all three?
If we follow the economic dynamics, it reveals a few stress points in the legal services market:
Law Firms Face a Squeeze:
Firms investing heavily in AI tools and transformation
Clients demanding lower rates and lower overall fees
Firms cannot easily pass through AI costs
The tension is especially acute for high-stakes work. Clients will want lower costs without quality reduction, but there is not yet any substitute for human expertise.
As a result, firms must choose between (a) cutting costs (e.g. junior lawyers and support staff), (b) win more work, (c) accept lower profit margins, (d) risk lowering the quailty of work, or (e) adopt a new business model.
In-House Teams Win Short-Term:
Gaining efficiency from AI tools
Reducing outside counsel spend
Handling more work with same headcount
In-house teams will benefit from cheap and automated legal services in the short term, but they run the risk that there is a ticking time bomb. Since a percentage of AI-generated legal work is going out the door without adequate human review, we are building latent risk in the system. When disputes arise or when market conditions shift, the quality of legal documents will be tested. It may be years before discover the real cost of today’s efficiency gains.
Consumers Gain Access:
Democratization of basic legal understanding
Significant cost savings
Consumers and small businesses win, and even though the work may be lower quality than those done by humans, given the lower stakes, it is likely a net positive.
Segmentation and Specialization
AI isn't yet good enough for the work that matters most, but it's already too good to ignore for everything else. The legal profession is caught in the middle of this transition, unsure whether to accelerate adoption or pump the brakes. We believe the answer depends on the type of legal work you are performing.
The future of legal AI likely involves greater segmentation and specialization:
Consumer tier: Dominated by general-purpose AI. We believe there is no viable business model for specialized legal AI here.
Business-as-usual tier: AI-automated workflows with spot-checking. Specialized tools for contract management, basic review, and template generation. The winners will be technology companies that offer the best integrations and automations.
Complex tier: AI may act as a research and drafting assistant, with heavy human oversight. Some call this the "centaur" model of human + AI working together. We believe this is where most legal professionals will operate in the short term.
Bet-the-farm tier: Primarily human-driven with AI in supporting roles. Work will be relationship-based. Premium pricing will persist.
What's clear is that we need better frameworks for understanding which tasks AI handles well, which require human judgment, and how to verify quality (or perhaps accept that we cannot manually review everything).
The market is sorting itself out slowly. Once you accept the type of legal work you are delivering, the winners will be those who design their services and quality controls for your tier of legal work.
