AI-assisted engagement letters: what to automate and what stays human
By the ChamberOne team 5 minute read
An engagement letter is both a contract and a first impression. It tells a client what the firm will do, for how much and on what terms. Much of it is the same for every client: names, addresses, matter numbers, standard clauses. Some of it is the most important thing the firm says all week. A good process automates the first kind and protects the second. This article draws the line and suggests a way to make the letter quick without making it careless.
Key points
- Automate facts already on the file, such as names, matter details and standard wording.
- A lawyer must decide scope, fee basis, limits on liability and who the client is.
- AI can suggest plain wording and spot inconsistencies, but should not choose what the letter promises.
- Have a second person check names and numbers against the file before sending.
- Click-to-sign is a simple electronic signature, so check whether your document needs more.
Why the letter deserves care
Many disputes between clients and firms begin with a mismatch about what was agreed. A clear letter is the best protection for both sides. It records the scope, the fee basis, who will do the work and what the firm will not do.
Write the letter for the client, not for the file. A reader with no legal training should be able to say, after one reading, what the firm will do, roughly what it will cost and what to do if they are unhappy. If a lawyer has to explain the letter in order for it to be understood, the letter has not done its job.
That is also why speed is dangerous. A letter generated and sent in a minute, with a scope paragraph copied from a different matter, can cause more trouble than a slow one.
Suppose a four-lawyer firm sends engagement letters for every new instruction. A busy Friday brings two similar matters, and the scope paragraph from the first is pasted into the second, which involves a different party and a narrower task. The client reads the letter, assumes the firm will do more than it intended, and the mismatch surfaces three months later in a fee dispute. Nothing about the template was wrong. The check was missing.
What can be filled in automatically
Anything that is a fact already on the file can be merged into the template. This removes retyping, and with it the errors that retyping brings.
- The client's name, address and contact details.
- The matter title, number and type.
- The names of the responsible lawyer and the team.
- The parties to the matter, as recorded.
- The date, and the standard wording that does not vary.
- The firm's complaints information and other fixed terms.
What a lawyer must decide
Some content is a judgment about this client and this matter. A template can hold a default, but a person must confirm or change it each time.
- The scope: what the firm will do, and what it will not.
- The fee basis and any estimate, cap or fixed fee, and what triggers a change.
- Any limit on liability.
- Who is the client, particularly where several people or companies are involved.
- Special terms for unusual risks, such as urgency or a foreign element.
- Anything the client has asked to change.
- Client name and contact details.
- Matter title, number and type.
- Responsible lawyer and team.
- Date and standard wording that does not vary.
- The scope: what the firm will and will not do.
- The fee basis and what triggers a change.
- Any limit on liability.
- Who the client is.
- Special terms for unusual risks.
Where AI can help, and where it should not
A tool can suggest a plain-language wording of a scope paragraph, tidy the language, check the letter for inconsistencies, or point out that a fee clause refers to a rate card the matter does not use. These are helpful second looks.
It should not decide the scope, set a limit of liability or choose what to omit. Those choices have legal and commercial consequences, and the lawyer who sends the letter is responsible for them. Treat any suggested wording as a draft and read it as if a trainee wrote it.
A useful check is to ask a tool to read the finished letter as the client would and list anything that seems unclear, then decide yourself whether to change it. Treat its comments as the opinion of a reader who has not seen the file, which is exactly what a client is.
A workable process
Here is a sequence a small firm can follow.
- 1
Open the matter and record the client, parties and matter type.
- 2
Generate the letter from the template for that matter type, with the matter's details filled in.
- 3
The responsible lawyer writes or confirms the scope and fee paragraphs by hand.
- 4
A second person checks names, numbers and cross-references against the file.
- 5
Send to the client with a short covering note in plain words, inviting questions.
- 6
File the signed version on the matter and diarise any review point.
Signing and its limits
Many firms now send the letter for electronic signature. A simple click-to-sign approach, in which the client types their name, is convenient and often adequate for engagement terms. It is not the same as a certificate-based or qualified electronic signature. Check the rules that apply to the document and to your jurisdiction, and use a more formal route where they call for one.
Whatever the method, keep the signed version and a record of when and how it was sent.
Think also about what happens when a client does not sign. Set a date for a reminder, and decide in advance whether the firm starts work before the letter is returned. That is a policy decision for partners, not something a system should settle quietly.
Keep templates current
Assign an owner to each template. Review them at set intervals and whenever professional rules, fee arrangements or the firm's insurance terms change. Retire old versions so that nobody uses one by habit. Check the guidance from your regulator or bar on what a client care or engagement document must cover, because that is not something a template can decide for you.
After each closed matter, ask the responsible lawyer whether the letter matched what actually happened. If scope changed, or a fee clause caused confusion, note it and improve the template. Letters improve fastest when the people who sent them tell you what they would change.
- 1Owner reviews each template at set intervals
- 2Review again when rules or terms change
- 3Retire old versions
- 4Ask after each closed matter if the letter matched
- 5Improve the template
- Then back to step 1.
How ChamberOne handles this
- You can draft from your own templates, with merge fields filled from the matter, the client and the parties.
- You can export to Word, or send the engagement letter to the client for signature.
- The client signs on their phone by typing their name on a page that shows only that document. You get a signed PDF with a certificate page, and the firm can counter-sign.
- The engagement letter template can come from the matter type.
What it does not do
- Click-to-sign is a simple electronic signature. It is not a certificate-based or qualified electronic signature.
- Some documents, such as deeds and wills, need formalities that an electronic signature does not meet. Check the rules for your document.
- ChamberOne does not give legal advice and does not decide the scope or fees in your letter.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
AI contract review: a first-pass checklist for lawyers
What to ask an AI to flag on a first read of a contract, what it tends to miss, and where a lawyer must read every word.
5 minute readWhat AI document analysis extracts well, and what it misses
What AI document analysis does well on parties, dates, clauses and citations, and what it misses, with ways to spot the gaps.
5 minute readClause libraries and AI drafting: keeping your own standards
AI drafts better when it starts from your vetted clauses, and worse when it starts from the open internet.
4 minute read