Legal
The hours a law firm cannot bill.
A dedicated AI model, trained on the firm’s own finished files and running on a server in its own office, does the work of opening a matter — so the people who bill by the hour are not the ones doing it.
A firm sells its people’s time, and the work of opening a matter is time nobody pays for. It is also the work that doubles the moment the matter count doubles, which is usually when a firm hires. The AI model reads what arrived, pulls out the detail, checks who has to be party to it, works out the costs and chases what is missing — and hands a lawyer a draft rather than a pile.
It drafts on the firm’s own hardware. A lawyer checks it and signs, and no client document leaves the office.
- Industry
- Legal
- Where it runs
- A server in your own office
- Working for you in
- About 8 to 12 weeks
- Investment
- About a junior’s salary, for a year
The challenge
A firm’s profit is the gap between the hours it bills and the hours it pays for. Opening a matter sits on the wrong side of that gap, and it is the part nobody can send an invoice for:
- Read what arrived and work out what it is
- Key the parties, the dates, the amounts and the terms
- Check who has to be party to it, and whether they can sign
- Work out the fees and the costs
- Chase whatever is missing or does not agree
- File the document and its annexures where they belong
Take on twice the matters and you do twice of all of it. That is usually the moment a firm hires — not because the legal work needs another lawyer, but because the operational work needs another pair of hands.
It is where errors get in
A date, an amount or a name keyed wrong, or a party nobody checked. Found late, it costs far more than the hour it saved.
It is where clients wait
The matter cannot move until the file is open and complete, so the client hears nothing while an assistant works through the pile.
It is paid at the wrong rate
When it falls to a lawyer, the firm is paying a lawyer’s hour for work that bills at nothing.
Clio’s 2024 Legal Trends Report puts up to 74% of a firm’s hourly billable tasks within reach of AI, which it works out as 66% of the average firm’s hourly billable work.
Our approach
We do not configure a product and we do not put a chatbot on the desktop for people to remember to open. We train an AI model on the firm’s own material and put it inside the work that is already happening.
- 1We go through your processes with you and find the repetitive cognitive tasks — the reading, the keying, the checking, the chasing — that are holding the work up.
- 2The AI model learns from your own material: your finished files, your precedents, your correspondence, the way your firm decides things. Nobody else has this one.
- 3It is built into the way you already work, so your people are not asked to use a new tool. The output lands where they already look for it.
- 4A person reviews what it flagged and signs off. Their corrections go back into the training, so each draft starts closer than the last.
There is no code in any of this, so there is nothing to re-code. No business analyst writing a spec every time something changes, no developer building it, no testing, no UAT, no release, and none of it again six months later. When the work changes we retrain the AI model on the new examples. It is also why a file that does not look like the last one gets a sensible attempt instead of an error, where a rule set would simply stop.
We train the AI model the way you would train a junior, by showing it the work. The difference is that you train it once, it does not resign, and when the business changes you retrain it rather than hire and start again. What it learns stays with you: the firm’s own way of working, held in an AI model you own.
What it handles
Works out what the document is
Contract, transfer, power of attorney, pleading, order, minute — identified, then the parties, the subject, the dates and the terms pulled out and written straight into the matter.
Checks who has to be party to it
Whether everyone who has to join in has joined in, and whether an attorney’s authority actually stretches to this transaction.
Works out the fees and costs
The duties, the registration fees and the adjustments that attach to the matter, under your firm’s own rules rather than a vendor’s.
Chases what is missing
Identification, signed forms, certificates, anything two documents disagree about — drafted as a request to the client, sent or held for a person, whichever you decide.
Files it
The document and its annexures, stored where the next system and the next person can reach them, rather than in somebody’s inbox.
It runs in your office, so the question of what leaves it never arises
For a firm, the problem with a cloud AI service is not the price. It is that putting a client’s file through somebody else’s system is a question you then have to answer — to the client, and about privilege. This one runs on your own hardware, so there is nothing to answer.
- No document leaves for processing. No upload, no outside service, nobody else holding or learning from your clients’ material.
- No new third party to disclose. Not a policy promise — there is simply nothing transmitted.
- No bill that grows with your volume. The AI model is yours and it runs on your server, so using it more does not cost more.
- The file and its trail stay inside your own systems, which is where your record-keeping obligations already sit.
The AI model drafts. Where a call needs experience — an ambiguous clause, an authority that does not quite reach, two documents that disagree — it marks it and waits rather than guessing. You decide, task by task, what goes straight through and what always stops for a person, and that line is set in the scoping.
The results
What the whole project costs, about a year of a junior’s pay, and it pays that back in a few months · indicative
The AI model absorbs exponentially more than one assistant’s routine load, so a busy month stops meaning another hire
From the scoping call to the AI model doing the work. What moves the date is how easily your systems connect, not the training and tuning of the AI model · indicative
What that buys is not a smaller firm. It is the same people, freed from the part of the day that bills at nothing, and a practice that can take on more matters without another pair of hands to open them. The figures above are indicative; on the scoping call we work them out again with your own numbers.
Common questions
Does anything leave the office?
No. The AI model runs on a server in your own office. There is no upload, no outside service holding your clients’ documents, and nothing for you to disclose to a client or assess as a new third party.
Does the AI model decide anything, or does a lawyer?
A lawyer decides. The AI model reads, extracts, checks and drafts, and where it is not confident it flags rather than fills in. Every matter is signed off by a person, and what they change is what trains it.
Our documents are not consistent. Is that a problem?
It is the normal starting point, and it is why the AI model is trained on your own finished files rather than configured against a template. It learns the variation. Where a document is genuinely unreadable — an old scan, a bad photocopy — it says so instead of guessing.
What changes for our assistants?
They stop doing the part of the day that bills at nothing and keep the part that needs them. This is built so a firm can take on more work without adding people at the same rate — it is not a redundancy tool, and it could not be: the corrections your people make are the only thing that keeps it improving.
How long before we see it working?
About 8 to 12 weeks to a working system, with the scoping before it. What moves that is not the legal work — it is how your systems are set up and how much has to be built to reach them.
Is your first stretch the same as this?
A scoping call is a conversation about which of your tasks are worth taking on, and what they are costing you now. It is free, and there is no obligation.
Prefer to talk? Call +61 475 043 070