Getting Started: The Horizontal On-Ramp

For most firms, AI rollouts begin with a safe first step through horizontal tools.

Leadership wants a proof point, so they start with something familiar - ChatGPT Enterprise, an internal wrapper, Harvey, Legora, maybe even CoCounsel or Copilot.

This phase is less about capability and more about getting their reps in with acquiring AI software. Firms figure out procurement, navigate infosec, and get through training and onboarding.

Everyone gets to try the tech, and the firm becomes smarter in its evaluation of tools. Once the firm has realized a level of comfort in seeking out and implementing general tools, the real work begins.

Getting Specific: Vertical Tools Take the Lead

The next wave of adoption is more targeted - this is where a lot of firms are spending most of their attention.

This is also where firms start looking at tools built for very specific jobs - ones that actually integrate with how a practice group works.

There are two potential paths here: adapt the horizontal tool for practice-specific workflows, or look towards vertical tools that are built only for a specific practice.

For a lot of practices, a tool like Harvey or Legora can easily coopted to fit the workflows. For example, their tabular views are a great fit for M&A due diligence.

But in the areas they serve, vertical tools tend to outperform the general-purpose products because they’re designed around legal workflows. Right now the most activity with vertical tooling has been in personal injury, with players like Supio and EvenUp raising serious rounds to own that niche.

Within IP, tools like DeepIP, Patlytics, and Solve Intelligence are carving out early traction.

Ultimately, when a tool is built for a workflow instead of a prompt, the output gets better, the firm takes notice, and adoption becomes that much deeper. Better performance within AI tools tends to lead to fewer review cycles, less second-guessing, and ultimately, more trust in the system. That trust is what drives real adoption across the firm.

Why Startups (Still) Matter to Lawyers

This shift toward specialization isn’t just about the tools, it’s also about the companies building them.

Most legal AI startups are making strategic bets:

Some go deep on a single practice area

Others go wide across firm types.

A company building patent automation, for example, has to serve firms of all sizes to be viable from a TAM perspective. Whereas a horizontal startup might target only enterprise firms or only SMBs. These choices shape everything from product design to support and pricing.

Understanding how a startup is built helps you understand whether it’s built for you. This makes it easier to evaluate which tools will actually work inside your firm.

What the Future Stack Looks Like

Every firm is building its own AI stack. What it looks like depends on size and practice areas.

Smaller firms that stick to one or two verticals will likely choose a single tool for the practice of law and another for the business of law. Simplicity wins.

For larger firms it looks more like a constellation. A few horizontal tools (ie: Harvey, Legora, Copilot) plus infrastructure tooling like DeepJudge and a data lake. Add PointOne for billing, a CRM, doc management, and then a growing list of vertical tools for each practice area.

Horizontal tools are a solid entry point, but they only go so far. The more specific the workflow, the more important it becomes to have tools that are built for that use case. This is where vertical tools tend to drive better performance, providing solutions that are much more contextual to specific problems within the firm, applying the solution only where it counts.

Answering The Ultimate Question

The “what are we doing about AI?” moment is turning into something more operational. The legal industry is building its AI stack in public with some firms going wide and some going deep.

The smart ones are asking: What tool actually works here? Not in theory, in practice.

As tools specialize and funding accelerates, expect the stack to evolve. It won’t be long before every firm looks more like a software portfolio than a software buyer.