Zendesk has no dictation for the reply composer. It has macros for canned text and AI that
drafts from your existing content, but nothing that turns speech into typing. Agent Workspace
runs in a browser, so on Windows this falls to voice typing
(Win + H)
or a dictation app. Steps below.
It's a browser tab. Zendesk lists the supported browsers; there's nothing else to install on your side.
2
Apply your macro first, if you're using one
Macros are built to fill the composer, so let the canned text land before you add anything of your own.
3
Click where the custom part goes
Public reply or internal note. Dictation types at the cursor, so put the cursor where you want the words.
4
Hold the key and talk
Ctrl + Space with Pithflow, Win + H with Windows voice typing. Explain it the way you'd explain it on a call.
5
Release the key
That ends the recording, and the cleaned-up text lands in the composer.
6
Read it, then submit
Nothing leaves the ticket until you choose a status and click submit.
Step 5 is the one that trips people up
Text does not appear while you're talking. The real order is this:
Hold
Ctrl + Space
›
Speak
in a normal voice
›
Release
recording ends
›
It lands
cleaned-up text
On a ticket that wait is the reread. The paragraph arrives finished instead of assembling
itself word by word with your false starts still in it, which is the version you'd have had
to clean up before submitting anyway.
Doesn't Zendesk AI already write replies for you?
It writes some of them, and it's worth knowing exactly which, because the answer decides
whether any of this is useful to you. Zendesk documents a set of AI features under
agent copilot, available as an add-on. The ones that touch the composer:
Suggested first replies — Zendesk describes these as suggesting a first response "based on existing macros and help center articles".
Enhance writing — expand, simplify, rewrite in a different tone, or run a custom prompt over a draft you've already got.
Suggested macros and similar tickets — surfacing the canned response or the resolved ticket that probably applies here.
Ticket summaries, intelligent triage and auto assist — context and next-step suggestions around the reply rather than inside it.
Zendesk also documents transcription and summarization for recorded phone calls, so there's
speech-to-text in the product — pointed at the customer's voice on a call, not at yours while
you write.
Read that list back and the pattern is clear: all of it generates text for you, out of
material you already have. Nothing in it captures what you would have said. That's the gap, and
it's a real one, because the tickets that eat an agent's day are exactly the ones no macro
covers and no article describes.
A reply is three different writing problems
Support writing gets talked about as one task, which is why tools keep getting compared as if
they're interchangeable. Split it properly and each piece has an obvious owner.
Canned text → macros win
The refund policy, the escalation notice, the shipping window. Zendesk applies a macro in
one click and it can set status, tags and assignee at the same time. Speaking text you
never change would be slower and worse. Use the macro.
Documented answers → AI drafts win
When the answer already lives in a help center article, generating from it beats retyping
it. That's what the suggested reply is good at, and it costs you a read-through rather
than a rewrite.
The bespoke paragraph → dictation wins
"Here's what actually happened on your account, here's why the first fix didn't hold, and
here's what I'm doing next." Nobody wrote that down in advance. It's yours, it's different
every time, and it's the part you retype all day.
Speak the paragraph the macro can't hold
Same key in the ticket, the CRM tab and the email client.
Two of those rows go the other way on purpose. Macros do something dictation cannot — one click
that writes the reply and moves the ticket — and voice typing is already sitting on the
machine with nothing to install. Neither of those goes away because you started dictating.
None of the three works offline: Microsoft documents voice typing as using online speech
recognition and requiring an internet connection, and Pithflow processes in the cloud. The
on-device option on Windows 11 is Voice Access, a separate feature.
What a spoken reply looks like once it lands
Explaining a fault out loud sounds nothing like a written reply, which is the objection people
raise before trying it. This is the kind of repair the AI cleanup does — an illustrative
example, not a measurement:
What you say
"ok so um the reason the first fix didn't hold is the invoice was still attached to the old
billing profile, i mean the one from before the account merge, so i've moved it over and
uh re-run it on our side, you should see the corrected invoice within the hour"
→
What gets typed
"The first fix didn't hold because the invoice was still attached to your old billing
profile, from before the account merge. I've moved it across and re-run it on our side —
you should see the corrected invoice within the hour."
Where Pithflow fits
There's nothing to connect, no marketplace app, and nothing to configure per field. Pithflow is
a Windows app that types into whatever field has focus, and because Agent Workspace is a
browser tab, every part of it qualifies: the public reply composer, an internal note, a ticket
field, the search bar. The same hotkey then carries into the CRM tab and the email client, so
an agent isn't learning one tool per window.
If your floor runs on virtual desktops, nothing installs inside the remote session — audio is
captured on the local machine — which is covered on
dictation in Citrix
and
dictation over Remote Desktop.
For how this looks across a shift rather than a single ticket, there's the
customer support use case.
Frequently asked questions
Does Zendesk have dictation for the reply composer?
Not for turning your voice into text as you write. Zendesk's AI features draft and rewrite text for you, and Zendesk documents call transcription and summarization for phone conversations, but there is no microphone button that types what you say into the reply box. On Windows that's a system-level job: voice typing (Win + H) or a dictation app.
Do I need to set anything up per field, or install a Zendesk app?
No. Agent Workspace runs in a browser, and Pithflow types into whatever field has focus. The public reply composer, an internal note, a ticket field, the search box — they're all just text fields to it. There's no marketplace app to add and nothing to configure in your Zendesk instance.
Should I use macros or dictation?
Both, for different halves of the reply. A macro is the better tool for text that never changes — Zendesk applies it in one click and it can set the ticket's status, tags and assignee at the same time, which dictation can't do. Dictation is for the part a macro can't hold: the paragraph about this specific customer's specific problem. Apply the macro, then speak the custom section.
Isn't this what Zendesk's AI copilot already does?
It solves a different problem. Per Zendesk's documentation, suggested first replies are generated from your existing macros and help center articles, and the writing tools expand, simplify or retone a draft. That's excellent when the answer is already written down somewhere. It isn't your voice, and it can't cover the case nobody has documented yet — which is the case that takes an agent the longest.
Does the reply send when I release the key?
No. Releasing the key ends the recording, and a second or two later the cleaned-up text lands at the cursor and stays there. Nothing submits. You still read it, pick your submit-as status, and click.
Does it work on a locked-down or virtual desktop?
Often, yes — because nothing installs inside the remote session. Audio is captured on the local Windows machine and the text is delivered into the focused field, so a Citrix, VMware Horizon or Remote Desktop session sees ordinary input. That's a common shape for support floors, and it's covered in more detail on the dictation in Citrix and remote desktop pages.
Can I dictate internal notes as well as public replies?
Yes, and internal notes are a good place to start, because the stakes are lower and the writing is looser. Same key, same behaviour — the text goes wherever the cursor is. Nothing about the composer changes what Pithflow does.
Does it work offline, and what does it cost?
It doesn't work offline: Pithflow processes in the cloud. Windows voice typing is online too — Microsoft documents it as using online speech recognition — and Zendesk itself needs a connection anyway. There's a permanent free tier of 2,000 words per week with no card, and Pro is $9.99 USD per month for full-shift volume. Full pricing is on the pricing page.
Want a different key?
Open Pithflow, go to Settings → Hotkey, click the
combination you want to replace, and press the new one. It takes effect immediately.
Hold to speak — Ctrl + Space by default. Hold it, talk, let go.
Hands-free — Ctrl + Shift by default. Tap to start, tap again to stop.
Worth changing if the default clashes with a shortcut you use. A spare function key such as
F9 is claimed by almost nothing.
Let the macro do the canned half
Speak the other half — in the ticket, the CRM, and every other text field on Windows.