Legal document automation is a template with variables. Client data goes in one end, a finished document comes out the other. No AI judgment calls, no hallucinated case citations, no surprises. It is the oldest form of legal automation and still the most reliable one, which is exactly why it’s the one my firms lean on hardest. My lemon law firm sends demand letters at volume, and most start life as a template that filled itself before a human ever opened the file.
If you automate one thing in your practice this year, make it a document.
Why templates beat generation for repeat documents
Generative AI writes new language every time you ask. That’s the right tool for a brief section or a first-draft contract clause. It’s the wrong tool for the retainer agreement you send 40 times a month, because you don’t want 40 slightly different retainer agreements. You want one agreement your malpractice carrier has seen, with 40 different names on it.
A template with variables gives you that. The logic is dumb on purpose: {client_name}, {vehicle_year}, {purchase_date}, merge, done. Dumb logic doesn’t hallucinate. When a demand letter goes out under my bar number, I want the persuasive language locked and only the facts changing. That’s the whole design principle behind how my firms handle demand letters.
The tools
Five names come up in every document automation conversation. Here’s how they actually differ:
| Tool | Price | Best for |
|---|---|---|
| Autocrat | Free | Google Workspace firms; merges Sheets data into Docs templates |
| Woodpecker | ~$39+/mo | Word-native firms that want templates living inside Word |
| Gavel (formerly Documate) | ~$99+/mo | Client-facing flows: client fills a form, document assembles itself |
| Clio Draft (formerly Lawyaw) | Priced with Clio | Court forms and firms already living in Clio |
| HotDocs | Enterprise | Legacy installs at large firms; powerful, dated, expensive |
Autocrat is the budget king. It’s a free Google Docs add-on that reads rows from a Google Sheet and produces one finished document per row. Zero legal branding, zero legal pricing. If your firm runs on Google Workspace, you can automate your first document today for nothing.
Woodpecker lives inside Microsoft Word as an add-in, which matters if your templates have years of formatting baked in that you don’t want to rebuild. Around $39 a month and your paralegals never leave the program they already know.
Gavel is the one to look at when the client should do the data entry. It builds intake questionnaires that feed directly into document assembly, so a client answering 20 questions on your website produces a completed document in your inbox. That client-facing layer is what you’re paying ~$99 and up for.
Clio Draft is the court forms play. It carries a library of fillable state court forms and syncs matter data from Clio, so contact and case details flow into forms without retyping. If Clio is already your system of record, this is the shortest path.
HotDocs has run document assembly at big firms since the 1990s. It’s deep and it’s proven, but the pricing and setup assume an IT department. Small firms should not start here.
The build-it-yourself alternative
Here’s what the tool vendors won’t tell you: the template creation step, the part that used to take a consultant a full day, now takes minutes. Paste any past document into Claude and ask it to replace every client-specific fact with a bracketed variable. It returns a clean template plus a list of the variables it found. That template drops straight into Autocrat or Woodpecker.
One step further: Claude Code builds the whole generator. Form on one side, Word document on the other, your logic in between. I built my lemon law demand letter generator this way. Intake fields go in, a formatted demand letter comes out with the firm’s language intact, and there’s no per-seat subscription because I own the code. My firms stopped renting document software the day this became possible, and the build took an evening, not a quarter. The full case for owning your tools is in Claude Code for lawyers.
Which one to pick
On Google Workspace and watching costs? Autocrat, free, today. Clients capable of filling out their own intake forms? Gavel, because the client-facing flow is the product. Living in Clio and filing court forms? Clio Draft. High volume with requirements no vendor quite fits? Build it with Claude Code and own it.
For contracts specifically, where clause logic gets conditional and the drafting is genuinely variable, template tools run out of road. That’s a different tool category and a different article.
Automate your first document this week
- Pick your highest-volume document. Not the most complicated one, the one you produce most often. Retainer agreements and engagement letters are the usual winners.
- Mark the variables. Open your last three versions side by side. Everything that changed between them becomes a bracketed variable. Claude does this pass in one prompt.
- Choose the tool. Use the decision guide above. When in doubt, start free with Autocrat and upgrade only when you hit a wall.
- Test with your last real matter’s data. Feed in the actual facts from a matter you already completed, then compare the output against the document you actually sent. Differences are template bugs. Fix them now, while the stakes are zero.
The quality rule
An attorney reads every generated document before it goes out. Every single one, until the template has survived 20 or 30 real matters without an edit. Then you drop to spot-checks. This isn’t AI paranoia, since a merge template can’t invent facts. It’s how you catch the variable you forgot to mark and the edge case your template never met, like the client with two vehicles or the hyphenated name that breaks your formatting.
Automation removes the typing. It doesn’t remove your name from the signature block.
Your move: before Friday, pull the document your firm produced most often last month and mark the variables. That’s an hour of work, and it’s the last time anyone at your firm builds that document by hand.