Document automation for small law offices without changing your templates

Every small law office has a shelf of documents that get produced again and again: powers of attorney, declarations, contracts where the base text is settled and only the parties, facts, dates and amounts change. Document automation promises to speed this up. The catch, in most tools, is the word "first". First, rebuild your template.

The dominant model: template as questionnaire

The established document automation products share one architecture. You convert your Word document into a tagged template, build a questionnaire, and the document is generated from the answers.

Gavel (formerly Documate) is a clear example of the model. In Gavel, a workflow is a questionnaire: you tag your Word template with variables in double curly braces, using their Document Tagger add-in for Word, build an intake questionnaire in a web builder, share the questionnaire link, and the document generates from the answers. Variable names are case-sensitive and must match the workflow. It is a capable product, with client intake portals, conditional logic and integrations, and its published model is questionnaire-driven: the data comes from a human answering questions.

MyCase Advanced Document Automation, the product formerly known as Woodpecker before its 2021 acquisition by MyCase, works inside Word as an add-in. You build reusable templates with fields, conditional logic, dropdowns and formulas, and they populate from MyCase client and case data. It is genuinely inside Word and deeply integrated with practice management, and it is sold as part of MyCase’s Advanced tier rather than standalone. Here too, the unit of work is a pre-built template with fields, filled from a structured database or by manual entry. We found no documented support in either product for filling from uploaded source documents such as scans or ID photos.

Why this model fits small offices badly

The questionnaire model is a good fit when documents are produced at volume from clean structured data, or when clients fill in their own intake forms. A two-to-five person law office usually looks different.

• The templates are settled. They have been refined over years, sometimes decades. Their wording, numbering and layout carry professional judgement, and in some practices institutional expectations too. Rebuilding them as tagged templates is real work, and every future change now has two places to be made.
• The data does not arrive structured. It arrives as a photographed ID, a scanned title deed, a PDF from a registry, an email from the client. Someone still has to read those materials and type the answers into the questionnaire. The retyping step has moved, not disappeared.
• There is no one to maintain the system. Tagged variables, case-sensitive names and conditional logic are a small internal software project. In a small office that project belongs to whoever is least busy, which means nobody.

The result is familiar: the office evaluates an automation tool, sees the rebuild cost, and goes back to copying last month’s document and overtyping the old client’s data. Which is its own error class, because a missed overtype leaves the previous client’s name in the new document.

The alternative: fill the document you already have

Turn the model around. Instead of restructuring the template to fit the tool, the tool reads your materials and fills your existing document.

• The template stays a Word document. No tags, no variables, no questionnaire. What you upload is what gets filled, with fonts, headers, spacing and layout unchanged.
• The materials stay in their native form. Photos, scans, PDFs, other documents, pasted text, dictated details. The software does the reading, not a person at a keyboard.
• The output is the same Word file you started from, with the changing values entered in the right places.

This is the model Run Done is built on. You upload the Word document you already use, add the materials you have, and the data lands in the right places. Nothing is invented and nothing is filled by the software’s own judgement: values come from the materials you supplied, and a value that cannot be found leaves the field empty. A document you produce often can be saved and filled again in seconds, and a task pane inside Microsoft Word runs the same flow in the document already open in front of you.

Where professional responsibility lives

A lawyer cannot delegate responsibility for a document to software, and should not want to. The question to ask of any automation is not whether it removes review but whether it structures review well.

• Review by exception: you check a list of entered values rather than re-reading a document you wrote years ago.
• Gaps stay visible: a value the materials do not contain is an empty field you can see, not a plausible guess you cannot.
• The final act is yours: you review, correct anything you want, and only then download. Control over the finished document never leaves the office.

The short version

Questionnaire-driven automation asks a small office to pay its scarcest resources, time and maintenance attention, up front. If your templates are settled and your data arrives as scans and emails, look for the opposite shape: a tool that reads the materials and fills the document you already trust, with your review as the last step.