Short answer: Dynamics 365 has no microphone button of its own. To dictate in it, click into any field and hold a system-wide hotkey on your Mac, or switch to a keyboard with a mic on iPhone. Nothing is installed in your tenant and no admin approval is needed.
Every CRM has the same disease, and Dynamics 365 has a particularly severe case of it: the records are complete on the fields that are dropdowns and empty on the fields that are sentences. Stage, close date, estimated revenue, owner, all populated. The note explaining what actually happened on the call, blank. Or worse: "Called. Followed up."
Managers usually read that as a discipline problem and respond with a discipline solution, which is a policy that says notes must be logged. Six weeks later the notes say "Called. Followed up." again, because the problem was never discipline.
It is an input problem. After a forty-minute customer call you can recount what happened in about ninety seconds of speech. Typing that same account takes six or seven minutes, and it arrives at the end of a call, right before the next one, when you have the least appetite for it. Faced with six minutes of typing or four words, most people pick four words. They would not pick four words if the honest version cost ninety seconds.
Dynamics 365 does not have a dictate button
Let us get the disappointing part out of the way. There is no microphone in the Dynamics 365 timeline note editor, no mic in the opportunity description, none in the case resolution box. Microsoft's dictation features live in specific products like Word and Outlook, and dictation in one of those is not dictation in your CRM. Different app, different text editor, different story.
There are also voice-adjacent capabilities in the Dynamics ecosystem that sound like the answer and are not. A conversation-intelligence feature that analyses call recordings is doing the archive job. It is not going to put your sentence into the field you are looking at right now.
What actually works is dictation that lives one level below the application. Voice Keyboard Pro types at your cursor at the operating-system level, so it does not need to know what Dynamics is. If a text box has focus and accepts typed characters, it accepts dictated ones.
Why this matters more in enterprise than anywhere else
If you have ever tried to get anything integrated into a Dynamics tenant, you know the shape of that process: a request, a security review, a data-handling questionnaire, a solution import, a sandbox test, a change window. Measured in quarters, not afternoons.
Nothing here touches your tenant. There is no solution to import, no connected app to approve, no admin consent screen, no entry in your environment's app list. It is an application on your own Mac that types. Your Dynamics deployment does not change in any way, and neither does the security posture your IT team already signed off on.
That said, whether you are permitted to use a third-party tool on a work machine with customer data on screen is your employer's call, not ours. Ask. We would rather you ask first than uninstall later.
Setting it up
On a Mac
- Install Voice Keyboard Pro and grant it the permissions macOS asks for. It lives in the menu bar.
- Pick a hotkey you can hold comfortably. Check it against the shortcuts you already use in your browser and in Dynamics itself, which has its own set of keyboard shortcuts for saving and navigating records. If your hotkey collides with one of them, change the hotkey rather than fighting it.
- Open Dynamics and click into the field first. This is the step people skip. Dictation types wherever the cursor is, so if the cursor is nowhere, the text goes nowhere. Click into the note box, see it focused, then hold the key.
- Speak a full sentence, then release. The text appears in the field, exactly as if you had typed it.
On an iPhone
Sales work happens standing up, in car parks, between meetings, which is where the mobile app matters. Add Voice Keyboard Pro as a keyboard in iOS Settings and enable Full Access so the microphone works. Then in the Dynamics 365 mobile app, tap into a field, press the globe key to switch keyboards, and use the mic.
The same flow works in mobile Safari if you use the web app on your phone. It works in the Field Service mobile app too, which matters if your notes are written next to equipment rather than at a desk.
The five surfaces worth dictating
1. Timeline notes and call logs
This is the main event and it is where most of the value is. The timeline note is the field that decides whether the next person to touch this account walks in informed or blind, and it is the field that gets four words.
Speak it in five beats and it comes out structured without you having to think about structure:
- Who was on the call and in what role. Names and titles, so the next person knows whether they spoke to a champion or a gatekeeper.
- What they said they need, in their words rather than your product's words.
- What is in the way. Budget timing, a competing project, a procurement process, an internal sceptic.
- What was agreed, including anything you promised to send.
- What happens next and by when, and whose move it is.
Spoken at a normal pace that is under two minutes. Typed, it is the thing you never do.
2. The opportunity description
Pipeline reviews go badly when the record has numbers and no narrative. Everyone can see the amount and the stage; nobody can see why this deal is at that stage. The description field is where the narrative belongs and it is almost always empty because nobody wants to type three paragraphs into a CRM.
Dictate it as if you were explaining the deal to a colleague who just joined: what they are trying to fix, why now, who signs, what would kill it. Refresh it after any call that changes the story. The point of a deal narrative is that it is current, and a narrative that is expensive to update stops being current within a week.
3. Emails sent from the record
Sending the follow-up from inside Dynamics is how it gets attached to the record instead of living in your personal sent folder. The problem is that composing in a CRM email editor is joyless, so people compose in their real mail client and the CRM loses the thread.
Dictating the body removes the reason to leave. Speak the follow-up while the call is still fresh, in the record, where it belongs. It will also be a better follow-up, because a spoken follow-up naturally references specifics from the conversation, while a typed one drifts toward the template you reuse for everyone.
4. Case notes and resolutions in Customer Service
Support records have the same failure mode, with sharper consequences. A resolution that says "fixed" teaches nobody anything. A resolution that says what the customer reported, what the actual cause turned out to be, what you changed, and what to check if it recurs, becomes the answer for the next person who gets the same ticket.
The reason resolutions are thin is not that agents do not know the answer. They know it perfectly at the moment they close the case. They are just out of typing budget at the end of a long interaction. Ninety seconds of speech is a budget most people do have.
5. Tasks, appointments, and account relationship notes
The small stuff adds up. Task descriptions that say what the task actually is, rather than "follow up." Appointment notes with an agenda in them before the meeting. Account-level notes about the relationship itself: who is new, who left, who prefers a call over an email, which internal team the account is politically attached to. All of that is institutional memory that lives in one person's head until it does not.
Speak the sentences, type the values
The single rule that determines whether this works well: dictate prose, type data.
Type these, always:
- Amounts, quantities, discounts, and forecast figures
- Dates that go in date fields
- Option set and choice field values, which are picked, not spoken
- Record IDs, quote numbers, case numbers, and anything with a GUID in it
- Email addresses, URLs, and account numbers
- Anything in a Power Fx formula or a business rule
The reason is not that dictation cannot produce digits. It is that number errors are silent. A misheard word in a sentence looks wrong immediately. A misheard digit in a revenue field looks perfectly plausible and flows into a forecast nobody re-checks. If it goes in a structured field, use the keyboard. If we wrote it out for you, the full method is in how to dictate numbers and dates, including when spoken numbers are safe and when they are not.
Two Dynamics-specific cautions
Save often. Long-form dictation into a browser form and session timeouts are natural enemies. If you are about to talk for three minutes into a note, save at the halfway point. For anything genuinely long, dictate into a scratch document first and paste it in.
Watch unsaved-form navigation. Dynamics will warn you about unsaved changes, but the surest way to lose four minutes of good notes is to click away from a form you were still speaking into. Finish the utterance, save, then navigate.
The proper-noun problem, and the fix
Here is what actually determines whether you keep using dictation in a CRM: the names. Account names, competitor names, product SKUs, the contact whose surname no transcription engine has ever spelled right. These words are maybe five per cent of what you say and they show up in one hundred per cent of your notes, so they set the emotional temperature of the whole experience.
Smart Vocabulary is a personal dictionary with replacement rules. You add the words you say constantly, and they come out right from then on. Load it once with your top thirty accounts, your product names, your competitors, and your internal acronyms, and the correction rate falls off a cliff.
This is the part that generic dictation cannot do. Built-in dictation on any platform never learns from your corrections. You can fix the same account name four hundred times and it will get it wrong on the four hundred and first. We wrote more about that specific mechanic in custom vocabulary that learns your words.
Where the notes should come from
The best CRM note is not written from memory an hour later; it is written from something you captured while the conversation was happening. On the Mac, Meeting Mode handles calls and meetings with speaker detection and AI notes, and calendar meeting detection means it knows a meeting is starting without you setting anything up.
That gives you raw material. Then you dictate the CRM note from it, which is a different and shorter task than reconstructing the call from nothing. Two things worth saying plainly: meeting notes are raw material, not a CRM record, and you should be transparent with the people on the call about capture. Consent norms vary by company and jurisdiction, and that is your organisation's policy call, not a decision a blog post should make for you.
If your team runs on a different system, the same approach applies. We have per-app versions for Salesforce and Zoho CRM, and a general playbook in dictation for CRM updates.
The habit that makes it stick
Log before you start the car.
Not at the end of the day, not on Friday, not in the batch you promise yourself you will do. The window where the call is still fully in your head is about ten minutes wide, and everything you log inside it is specific. Everything you log outside it is generic, because you are reconstructing rather than recalling.
With typing, that habit is unrealistic; nobody is opening a laptop in a car park to type six paragraphs. With a phone in your hand and a mic on the keyboard, it is ninety seconds. That is the entire behaviour change, and it is why the notes get better rather than just faster.
Fixing text on a phone without the thumb rodeo
Dictating on mobile is easy. Correcting on mobile is the miserable part: placing a cursor precisely in the middle of a paragraph on a touchscreen is a small act of violence. Voice Edit lets you speak the change instead. Say what should be different and the text updates, without you dragging a magnifier around.
For international pipelines, two-way translation while dictating covers 24 languages, so you can speak in your language and send in theirs. Treat machine translation as a strong first draft rather than a final one for anything contractual, and have a person check anything that carries legal weight.
The tone trap
One warning, because it costs people real deals. Typing is slow, and that slowness is accidental cooling time. By the time you have typed four sentences about a customer who wasted your afternoon, you have already softened three of them.
Speech has no such delay. Dictate a note or a reply while irritated and what lands in the record is exactly how irritated you were, in a system your colleagues and possibly your customer's account team can read. The rule is simple: never send or save a hot one in the same sitting. Speak it, walk away for a few minutes, read it back, then decide.
Frequently asked questions
Does this work with Business Central, Field Service, and the other Dynamics apps?
Yes. Because the text goes in at the cursor, the specific module does not matter. If it is a text field in a browser or an iOS app, you can dictate into it. That includes the model-driven apps your team has customised, which is usually where the interesting long-text fields live.
Do I need my administrator to approve anything?
Nothing is installed into your Dynamics environment, so there is no solution import or admin consent step. Whether your organisation permits third-party software on a managed Mac is a separate question and one for your IT team. If the machine is managed, ask before installing.
Is the customer data I dictate stored anywhere?
Our server stores only operational pings. No audio, no transcript content. Your dictation history stays on your own machine. For regulated data specifically, your compliance team owns that decision, not us, and they will want to read the policy rather than a summary of it.
Can I dictate into the Dynamics rich text editor?
Yes. Rich text fields accept dictated text the same way they accept typed text. Formatting like bold and bullet lists is still applied by the editor's own controls, so speak the words and use the toolbar for structure.
What about Copilot in Dynamics?
Different job. Copilot generates or summarises text for you; dictation captures what you actually want to say. If you use both, dictation is how you feed it a properly detailed prompt instead of a lazy one, which is generally where the quality of the output comes from.
Will my colleagues be able to tell I dictated?
Only in the sense that your notes will be longer and more specific than theirs.
What it costs
There is a free tier with daily limits, which is enough to test this on a week of real calls. Pro is $4.99 a month or $34.99 a year, and covers both the Mac app and the iPhone keyboard.
The honest limits
Dictation will not fix a CRM whose fields nobody agreed on, and it will not make a bad note good. A long empty note is worse than a short empty one, so the discipline of saying something specific still belongs to you. Structured data entry stays keyboard work. And if your objection to logging is that the process itself is broken, faster typing of the wrong thing does not help.
What it fixes is narrower and more useful than that: it removes the cost that turns a useful note into four words. Voice Keyboard Pro has a free tier. Try it on the next call you finish, before you start the car.