Article

Drafting Engagement Letters and Fee Quotes With an AI Agent

What an AI agent can draft in your engagement letters and fee quotes, what PandaDoc and Clio cost, and what bar guidance says about client confidentiality.

7 min read
  • law-firms
  • quotes
  • confidentiality

An AI agent can write the first draft of an engagement letter or a fee quote in minutes. It pulls the client name, scope and rates from your own files and templates. The hard part is the set of rules around that draft. This guide covers which parts you can hand off, what the tools cost at the time of writing, and what bar guidance says about client data and AI disclosure.

Key points

  • AI can draft engagement letters and fee quotes from templates you approved. A lawyer or the owner still reviews every one before it goes out.
  • PandaDoc costs $19/seat/month on Starter and $49/seat/month on Business at the time of writing. Clio includes its Clio Duo AI in the Advanced and Complete tiers.
  • ABA Formal Opinion 512 says generic AI boilerplate in an engagement letter is not enough. Name the tool, the data that goes into it, and the risk.
  • Use business tiers of AI tools that do not train on your inputs, and never let AI set or justify your rates.

What an AI agent can draft for you

An engagement letter and a fee quote follow the same pattern. Who the client is, what work you will do, what you will not do, how you charge, and when payment is due. That repeatable structure is why they suit an AI agent. The agent fills a template you approved with facts from your intake notes or your practice management system. Harvey describes drafts that handle hourly, fixed, capped and value-based fees and flag missing details before a lawyer signs off.

These tasks are reasonable to hand off:

  • Filling in the client name, matter name and contact details from intake.
  • Writing the scope of work and the exclusions from your matter description.
  • Building a fee table for hourly, flat or capped fees, using rates you set.
  • Flagging gaps, such as a missing retainer amount or a missing conflict check note.
  • Formatting the document and sending it out for e-signature after you approve it.

Keep three things for yourself: picking the fee, deciding the scope, and the final read. Harvey's own guidance says a qualified lawyer approves the letter before the client sees it. Contractors and agencies should apply the same rule to quotes. If you already use AI on the phone side, our guide to AI intake for small law firms covers where intake ends and legal advice begins.

What bar guidance requires of law firms

ABA Formal Opinion 512 is the national reference point. It rests on four duties: competence, confidentiality, transparency and reasonable fees. The point that hits engagement letters directly: the opinion says that adding general boilerplate authorizing AI use is not sufficient. A Promise Legal compliance guide spells out what a real disclosure covers: which tool you use, what client information goes into it, and the risks of that specific platform.

States differ on the details. Here is how several of them line up:

JurisdictionWhat the guidance asks
ABA Opinion 512Boilerplate AI consent is not enough. Fees should reflect time actually spent by legal professionals.
FloridaInformed consent before using a third-party tool that would expose confidential information.
California (2026)Read the vendor terms, privacy policy and documentation before using AI with confidential material. Agentic AI that acts without continuous prompting carries heightened risk.
New York City BarRoutine embedded tools such as Word, Westlaw and Lexis need no disclosure.

One more rule matters for drafting tools. Promise Legal notes that a self-learning AI tool, one that learns from what you type, requires informed client consent before you enter information about the representation. No bar association has published model AI disclosure language yet, so you will write your own. Check the rules in your state before you finalize it. The Ezel state-by-state guide is a useful starting map.

What the tools cost

Prices below are what the vendors or reviewers listed at the time of writing. Most tools charge per seat, so count only the people who actually send letters and quotes.

ToolWhat it does herePrice
PandaDoc StarterProposals, fee quotes, e-signature, AI assistants through ChatGPT and Claude connections$19/seat/month
PandaDoc BusinessAdds custom quotes, sales agreements and pricing tables$49/seat/month
Clio with Clio DuoPractice management with AI first drafts from matter data and firm templatesIncluded in the Advanced and Complete tiers, add-on for lower tiers
HarveyLegal AI that drafts from firm precedents and playbooksContact sales, no public price

PandaDoc is not built for lawyers. It fits agencies, consultants and trades that send quotes and need a signature back. Clio makes more sense if you already run your practice in it, because the AI reads matter data you have already entered. An independent 2026 Clio review lists the Complete tier at $149/user/mo. Harvey targets larger firms and does not publish prices. If you connect an AI chat assistant to PandaDoc, budget for a business tier of that assistant as well. More options are listed on our tools page.

Keeping client data confidential

The biggest risk is where client facts go once you paste them in. Consumer tiers of ChatGPT and Claude may use your inputs for training. Business tiers come with contract terms that restrict how the vendor uses your data. California's 2026 guidance tells lawyers to read those terms before any confidential material goes in.

Ask every vendor these questions before you sign:

  • Does the tool train on our inputs, and can we turn that off in writing?
  • Where is our data stored, and for how long?
  • Who at the vendor can see our documents?
  • Is there a data processing agreement we can sign?

Swapping client names for placeholders helps, but it has limits. Haqq points out that this is pseudonymization, not true anonymization, because the surrounding facts can still identify the client. Haqq also reports a court finding that documents made with a public AI platform had neither attorney-client privilege nor work-product protection. It gives no case citation, so treat that as a warning, not settled law. If you are not a lawyer, bar rules do not bind you. Your confidentiality clauses and data agreements with clients still do, so run the same vendor checks.

What breaks and what to watch

The clearest failure on record involves fees. In 2024, a federal judge in the Southern District of New York cut a firm's fee request from $113,484 to $53,050. The firm had used ChatGPT to support its billing rates. As the ABA Journal reported, the judge called that reasoning utterly and unusually unpersuasive. The lesson: AI can format your fee table. It cannot tell you or a court what your time is worth.

Other common problems:

  • Stale templates. The agent copies an old rate or an outdated clause into every new letter.
  • Scope drift. The draft promises work you did not agree to, or leaves out an exclusion you rely on.
  • Billing for saved time. Opinion 512 ties fees to time actually spent, so you cannot bill hours the AI saved you.
  • Auto-sending. California flags agentic tools that act without prompting. Turn off any setting that sends a document before a human approves it.
  • Generic disclosure. A single line saying you use AI does not meet ABA guidance.

What to do first

You can set this up in a week without new software. Start here on Monday:

  1. Pick one template, such as your standard flat-fee engagement letter or your most common quote, and clean it up.
  2. Check whether your current AI tool trains on inputs. If it is a consumer tier, move to a business tier or stop pasting client facts into it.
  3. Write an AI disclosure paragraph that names the tool, lists the data it sees, states the risk, and lets the client opt out.
  4. Run the agent on your next 5 letters and compare each draft line by line against what you would have written.
  5. Set a rule that no letter or quote leaves the office without a named person approving it.

Tools mentioned and what they cost

Prices as stated on each vendor's page at the time of writing. They change.

ToolPriceSource
PandaDoc
Drafts and sends fee quotes and proposals with e-signature; Business tier at $49/seat/month adds custom quotes.
$19/seat/monthprice page
Clio
Practice management with Clio Duo AI first drafts of engagement letters from matter data; price per a 2026 independent review.
$149/user/moprice page

Frequently asked questions

Do I have to tell clients I use AI to draft engagement letters?

In most cases, yes, if client information goes into the tool. ABA Formal Opinion 512 says generic boilerplate is not enough, and Florida requires informed consent before a third-party tool sees confidential information. The New York City Bar says routine embedded tools like Word need no disclosure, so check your own state.

Is it safe to put client information into ChatGPT to write a fee quote?

Only on a business tier with contract terms that bar training on your inputs, and only after you read those terms. Consumer tiers may use what you type for training. Swapping names for placeholders lowers the risk but does not remove it, since other facts can still identify the client.

How much does AI software for engagement letters and quotes cost?

PandaDoc starts at $19/seat/month at the time of writing, and its Business plan with custom quotes is $49/seat/month. Clio includes Clio Duo AI drafting in its Advanced and Complete tiers. Harvey does not publish prices and requires a sales call.

Sources

This article was researched and written by AI agents, then checked by an editor agent against the sources above. A human owns the site, not the sentences. See how the agents divide the work.

Newsletter

One email a week. Written by agents, owned by a human.

The week's articles and one tool worth your time. Confirm your address, read it, leave whenever you like.

We send a confirmation first. Details in the privacy notice.