Read this first

First, answer this question

If the answer is “faster than typing,” you need speech-to-text. That is a solved, inexpensive problem in 2026 and you should not spend much on it.

If the answer is “someone else finishes it” — a secretary, a typist, a paralegal formatting it into the firm’s template — then transcription is only half of what you’re buying. The other half is the document assembly, and that is where the interesting products are and where the cost sits.

Firms that answer the second way and buy for the first end up disappointed. Firms that answer the first way and buy for the second overspend by an order of magnitude.

1. Stay on Dragon — move to the desktop version

The appeal: same engine, same vocabulary, no retraining, and you own it outright.

The cost nobody mentions: you are reversing an architectural decision you made deliberately. You chose Anywhere because it works from home and from court, because a profile follows the user instead of the machine, and because IT stopped hand-maintaining installs on twelve desktops. Desktop perpetual gives all of that back.

Also worth knowing: desktop Dragon has its own retirement history. “Reported to continue” is not a commitment, and we would want a straight answer about its roadmap before signing.

2. Straightforward speech-to-text

The appeal: if the honest answer to the question above was “faster than typing,” this is genuinely sufficient, and modern general-purpose recognition is far better than the reputation it carries from a decade ago. Some are free. Most of the paid ones are a few dollars per user per month.

The limitation: general vocabulary. Case citations, Latin, and party names will need correcting. No document structure — you get text, and formatting is on you. No workflow, no secretary handoff, no template.

3. Legal AI platforms

The appeal: these are serious products doing serious work — contract analysis, review, research, drafting at a level well beyond a dictation tool.

The reality check: they are built and priced for firms with an innovation budget and someone whose job title includes “legal operations.” Pricing is generally not published, which is itself informative. Most are not voice-first — you type or paste; they are not designed around a recorded dictation as the input.

4. Microsoft 365 Copilot

The appeal: it drafts in Word, it knows your documents, and there is no new vendor to onboard. If your firm is on Microsoft 365 you may be able to test this next week at no additional cost.

The limitation, and it is the specific one that matters here: it does not take a recorded dictation as input. You type into the chat, or you dictate into the chat in real time. If your workflow is record now, produce the document later — walking out of a client meeting, dictating in the car, handing a file to a secretary — that workflow does not survive the move.

5. A modern cloud legal documentation platform

What it is: the successor generation to what Dragon Legal Anywhere was — cloud, subscription, dictation-first — but where the output is a structured draft in your template rather than a wall of unformatted text. You speak naturally, out of order, the way people actually talk. The system assembles the document. Better implementations let you attach source files so facts you forgot to dictate get pulled in, and keep the secretary or typist in the loop as a review step rather than removing them.

The appeal: it preserves the architecture you already chose — mobility, no desktop maintenance, profile follows the user — and it closes the gap that made Dragon Legal Anywhere only half a solution: you got text, and a human still had to turn it into a document.

Pricing: this category generally runs a few hundred dollars per user per year, which for most firms lands in the same range as what they already pay.

A rough guide

That last row is a real answer, not filler. If your renewal date allows it, extending inside 2026 gives you a supported product through 2027 and removes the time pressure entirely. A good decision in March beats a fast one in December.

What we’d tell you over the phone

There isn’t a single right answer here, and any page that ends with one is selling.

What we’d actually say is: the mistake to avoid is replacing like-for-like without asking what you were really buying. Dragon Legal Anywhere transcribed. It didn’t assemble documents. If your firm has been absorbing that gap with someone’s time, replacing it with another transcription tool re-buys the gap.

And the mistake we’d want to avoid making with you: recommending something more expensive than the problem requires. If Copilot is already in your stack and your attorneys draft at their desks, test it first. We would rather have the conversation after you have.

We work with firms on this. We are not the right answer for all of them.

Related reading