How an 18-Person Law Firm Recovered $1.3M a Year With Two AI Employees

An 18-person family law firm on the Sunshine Coast was drowning in 4,000 emails a month. Today, two AI systems handle that filing automatically and turn every client consultation into a polished letter in minutes. The result: the equivalent of $1.3M a year in recovered productivity, without the team having to change a single process.

This is the full story of what we built for Catton and Tondelstrand - what the systems do, how we rolled them out, and what actually changed. No hypotheticals. No "up to" figures. Real numbers from a real firm.

The problem: admin was eating the billable day

Catton and Tondelstrand is a busy family law practice. Like most firms their size, the bottleneck was not the legal work - it was everything around it.

Every email had to be read, categorised, and filed to the right matter in Smokeball. With 4,000 emails landing every month across 18 active users, that is hours disappearing from every person's day. Then there were the consultations: after each one, a lawyer or paralegal spent 30 to 45 minutes writing up file notes and client letters that followed the same structure every time.

None of it was hard. All of it was necessary. And all of it was pulling qualified people away from the work clients actually pay for.

Meet Pat: the AI email employee

Pat is the firm's AI email filing system. Every email that comes in gets read, matched to the correct matter, and filed into Smokeball automatically - no human dragging and dropping, no "I'll file it later" pile.

In March 2026 alone, Pat processed 3,844 emails. That is thousands of small decisions - which matter does this belong to? - made instantly and correctly, all day, every day, across the whole team.

The way to think about Pat is exactly how we describe AI to every client: it is a junior employee you just hired. Clever, tireless, and very good at one well-defined job once you have shown it how the firm works.

Meet Donna: the AI documentation employee

Donna handles the after-meeting admin. She transcribes client consultations and generates file notes and client letters automatically, in the firm's own tone and format.

What used to take 30 to 45 minutes of a fee earner's time after every meeting now takes minutes. The lawyer reviews and signs off rather than writing from a blank page. There is still a human in the loop - they just start at 80%, not zero.

They describe it as having an extra employee - one that works 24/7 and never calls in sick.

How we rolled it out: start small, build trust

We did not walk in and automate everything on day one. That is the fastest way to break trust and break a process.

We started with one thing: email filing. We proved it worked. We earned the team's confidence. Only then did we add meeting documentation. Start small, build trust, then build bigger - it is the same approach we take with every firm, because AI amplifies whatever process it is pointed at. Point it at a solid workflow and you amplify success. Point it at chaos and you just get faster chaos.

Every system is fully managed. We run it, optimise it, and fix what breaks - model updates, edge cases, the lot. The firm got AI employees without having to manage them.

The numbers behind the $1.3M

It is built from the two systems, measured on real usage:

• Pat (email filing): around $81,000 a month in recovered time across 18 users, based on the hours previously spent reading and filing 4,000+ emails.
• Donna (documentation): around $35,000 a month in recovered time, based on the consultation write-ups now handled automatically.

Together that is roughly $116,000 a month, or about $1.3M a year in productive time given back to the firm. We measured it on logins, email volume, and meeting counts.

Curious what the same maths looks like for your firm? Try our AI Savings Calculator.

Could this work for your firm?

If your team is filing emails by hand, writing the same letters from scratch, or losing the day to admin that follows a predictable pattern, then yes - almost certainly.

The starting point is never "buy the AI." It is a simple question: what takes you 30 to 45 minutes every day that nobody wants to do? That is where AI earns its keep first. Prove it on one task, then expand.

AI is not a silver bullet, and most tools out there are just ChatGPT with lipstick. But pointed at the right repetitive, pattern-based work - and built into how your firm actually operates - it changes the game. Catton and Tondelstrand are the proof.

If you are a law firm thinking about AI, the real question is not whether you can afford it. It is whether you can afford to keep paying qualified people to do work a system could do for them.