AI for attorneys: a decision framework for your first deployment

If you run a solo or small firm and you’ve been wondering whether AI is finally worth deploying in a law practice — not at BigLaw, where there’s an IT department and a vendor contract to sign, but at your practice — this is the post I wish existed when I started thinking through the same question for my clients.

I’ve deployed AI phone systems for attorneys. The question I get asked first is almost always the wrong one. “What AI should I use?” is not the right starting point. The right question is: “What’s actually costing me money right now, and can AI reliably fix that?”

The problem most attorneys have is the phone, not the billable hours#

Solo and small-firm attorneys already know how to do the legal work. The bottleneck isn’t research or drafting — it’s the operational drag that surrounds the practice. Intake calls coming in during depositions. Prospects leaving voicemail at 9pm on a Friday. Clients calling for case status updates that require you to look something up and call back before the day is over.

A study of 1,200 calls placed to law firms found that 35% went completely unanswered. After hours, that number climbs toward 90%. And research consistently finds that 67% of legal clients hire the first attorney who responds to them — not the best-reviewed, not the lowest-priced. The first one who picked up.

That’s not a marketing problem. It’s an ops problem. And it’s the one AI is actually positioned to fix right now.

What AI can handle for a law firm today#

Three categories make sense for AI phone or messaging handling at a solo or small firm. They’re not equally good fits, so here’s how I’d rank them.

Intake screening is the strongest case. A caller says they were in a car accident. Your AI asks: Were you injured? Have you already spoken with another attorney? What state did this happen in? It collects structured intake answers, tells the caller you’ll follow up within your stated window, and sends you a summary — by text, email, or directly into your CRM. You never have to personally screen that call. If your intake questions are well-defined (and most are), this is reliable today.

Appointment scheduling is also a strong fit. The AI books a consult directly into your calendar based on your availability. No email back-and-forth, no one at a desk doing the coordination. If you use Calendly, Acuity, or a similar tool, the AI can handle the handoff cleanly.

Case status questions need a caveat. If the information is structured — your next hearing is April 10th, your documents were filed on April 7th — AI can answer routine status questions against a knowledge base you maintain. But if the client’s questions involve judgment calls about strategy or timeline, route those to voicemail or live pickup. Don’t ask AI to improvise on case details.

What to leave alone for now#

Document review, legal research, drafting — AI can assist with these, but that’s a different tool category entirely. The AI Receptionist I deploy handles the operational layer: call answering, intake capture, scheduling. Not substantive legal work. Keep those separate.

The decision framework: pick one thing first#

You don’t need to automate everything at once. Pick the one workflow that’s costing you the most money or time and start there.

Here’s how I’d prioritize for a solo or small firm:

Start with intake if you’re missing calls. This is where the revenue leakage shows up most clearly. Every intake call you miss and don’t recover quickly is a case you don’t get. If your phone goes to voicemail after 5pm and you have any meaningful case volume, this is your first deployment.

Start with status updates if callbacks are draining your day. If you’re spending time each afternoon returning calls from existing clients asking where their case stands, build a simple knowledge base the AI can pull from. Clients call, AI reads them the relevant status from a shared doc or CRM entry. That alone can cut a significant portion of your callback queue.

Start with scheduling if consultation booking is the friction. Someone calls after hours, AI offers available slots and books the consult directly. You wake up with appointments on your calendar you didn’t manually schedule.

Rank these by where you’re bleeding time or missing revenue. For most solo attorneys I talk to, it’s intake — missed calls from prospects who didn’t leave a voicemail, or left one and called someone else by morning.

What this costs compared to hiring#

A legal intake coordinator in the U.S. earns around $39,000–$43,000 in base salary. Fully loaded with benefits, payroll taxes, and paid time off, you’re closer to $50,000–$57,000 per year. And that’s for business hours only — they’re not answering the phone at 8pm when a potential client Googles “DUI attorney” from their car.

A legal virtual assistant on retainer runs $2,200–$3,500 per month for part-time coverage. Still not around the clock, and still a recurring line item every month indefinitely.

The AI Receptionist I deploy is $8,000 one-time. After that, the monthly cost to run it is effectively zero — no SaaS subscription, no seat license, no per-call fee going to a third party. You own the deployment. If you want to see the full year-one math laid out side-by-side, the cost breakdown against a hire is here.

That math is why this product makes the most sense for attorneys who already have enough case volume to feel the intake problem, but not enough volume to justify a full-time staff hire just for phones.

When this isn’t the right move yet#

There are real situations where I’d tell an attorney to hold off.

Your intake has no consistent questions. If every new matter is highly variable and the initial call genuinely requires attorney judgment to evaluate — not just information collection — scripting it for AI is harder. You’d spend more time building the intake flow than you’d save in the first year.

You’re already at capacity. If your intake line rings three times a week and you’re booked out three months, AI for intake doesn’t move the needle. Spend that budget on something that actually expands capacity or cuts admin time elsewhere.

Your jurisdiction has specific bar rules about initial client contact. Some state bars have requirements around what can and can’t be communicated during intake. This is solvable with the right script review, but it’s a design requirement — not something I’d deploy without your sign-off on the intake language.

You haven’t decided what you want AI to say. This one isn’t attorney-specific — it’s the most common failure mode I see across all deployments. AI follows instructions precisely. If you haven’t mapped out the intake flow, the AI doesn’t invent it. Every client I’ve worked with who’s been disappointed came in saying “just set something up.” That’s not how this works.

The practical next step#

If you run a solo or small practice, the intake phone problem is real, it’s quantifiable, and it’s solvable without adding a new hire. The AI Receptionist isn’t magic — it’s a well-built phone system that handles the routine so you can focus on the billable work.

If you want to see what the deployment looks like specifically for a law practice — the intake flow, the calendar integration, the case-status knowledge base, and where the limits are — I’ve laid it out in the attorney-specific receptionist deployment. That page covers the exact workflow, what it handles from day one, and what it doesn’t.