How to Evaluate Legal Technology and AI

Most firms don't have a technology problem. They have a diagnosis problem.

We see a new tool. We sit through a demo. We feel modern for a minute. And then the tool sits there, solving a problem nobody actually had.

The firms that get the most out of technology rarely buy first. They ask better questions first: what kind of technology is this, what problem am I solving, and does it hold up under scrutiny?

Where I learned to ask questions first

I came to this industry through my grandmother's estate. (I've written about that before.) A "simple" estate, with no disputes and no complicated assets, took three and a half years to close.

Not because anyone was incompetent. The friction was built into the process: handoffs, waiting, and systems never designed to help anyone finish.

So my cousin Ari and I got curious. We talked to lawyers, accountants, trust officers, and families. Our experience wasn't unusual at all. Estate planning was crowded with new tools. Estate administration for professionals was nearly empty. And the same three pain points kept coming up:

• Court applications
• Task management
• Accounting‍

That became the foundation of Estateably. We didn't start with a technology looking for a use case. We started with a real, shared problem.

The anatomy of a good technology decision

1) Not all technology is the same
(Digitalization, Automation, and AI)

"Legal tech" gets used like one big bucket. It isn't.

• Digitalization: getting information into a format computers can work with. Paper to PDFs. Filing cabinets to searchable databases.
• Automation: fixed rules, no judgment. Same input, same output, every time. Think auto-calculated distributions.
• AI: judgment-based output that handles ambiguity. Think first-pass drafting, or flagging anomalies in an estate filing.

Put simply: automation replaces repetitive human action. AI replaces some human judgment. And the fanciest tool isn't automatically the right one. If your files are still on paper, AI won't fix that.

2) Identify the problem before choosing a solution
(Volume, Complexity, or Coordination)

This is the step most firms skip. Before I look at a single tool, I ask three questions:

1. Is this a volume, complexity, or coordination problem?
   Too many of the same task, a task that needs judgment, or a delay in people and handoffs?
2. Where does the friction actually happen?
   Pain is usually felt at data entry but often starts upstream, at intake or approval.
3. Would a process change fix it?
   If a checklist or template resolves 80% of the pain, do that first. Then automate a clean process, not a broken one.

The answer usually points the way: volume calls for digitizing or automating, complexity for AI with human review, and coordination for fixing the process before any tool.

3) Use a four-lens evaluation framework
(Before anything gets signed)

Every candidate tool goes through four lenses. This isn't a scorecard you average. Fail one, and it isn't safe for a fiduciary practice.

• Discovery: build your shortlist from AI-assisted search, traditional search, and your professional network, not whichever name is loudest.
• Accuracy and professional responsibility: test it on real work, and check it against your duties of competence, confidentiality, and supervision.
• Data security and integration: know where client data goes and whether the tool fits your systems before you sign.
• Business value: name the expected gain up front, so you have a benchmark to measure against later.

We hold ourselves to the same standard. When we chose a company-wide AI tool, the case came from people already getting better results on real work, not from a mandate.

The takeaway

If you're feeling pressure to "do something" about AI, start smaller than the market wants you to:

• Know which kind of technology you need
• Name the problem before you look at a product
• Run every option through all four lenses

Because the goal isn't the newest tool. It's work that moves, decisions you can stand behind, and more time for the people on both sides of the estate.

-----

Frequently asked questions

How should a trust and estate practice evaluate legal technology?

Start by naming the problem and classifying it as volume, complexity, or coordination. Then test each candidate tool through four lenses: discovery, accuracy and professional responsibility, data security and integration, and business value. A tool must pass all four.

When should a law firm use AI instead of automation?

Use AI for complexity problems, where the task requires judgment, and always with human review. Use digitization or automation for volume problems. For coordination problems, fix the process before adding any tool.

Should a firm change its process before buying new software?

Often, yes. If a checklist, template, or reassignment resolves most of the pain, do that first. Automating a broken process only makes the problem run faster.