Place the cursor
Click exactly where the final text should appear. DictaFlow returns to that active field after processing.
DictaFlow Medical quick start
Choose your device, complete one short practice dictation, and learn the few settings that matter most. Most people can be ready in under ten minutes.
For clinic pilots: sign in with the exact email your administrator approved.
Place the cursorOpen the EHR field, document, or message where the text belongs.
Start and speakUse your shortcut or the app control, then dictate naturally.
Stop and reviewYour text appears ready for clinical review.
Acute bronchitis. Continue albuterol every four hours as needed. Follow up in one week.
Always review before saving or signing.Choose your device
The guide switches automatically when it can. Choose another platform at any time.
Windows 10 or 11
Windows is the most direct desktop workflow: open a field, use the hold-to-talk shortcut, then release to insert the finished text.
Install DictaFlow Medical from the Microsoft Store. If both DictaFlow apps appear in search, choose the one with “Medical” in its name.
Use Apple, Google, or Microsoft with the exact email linked to your Medical account. Clinic-pilot access activates automatically after that first sign-in.
Allow microphone access when Windows asks. In DictaFlow Settings, choose the headset or microphone you plan to use and run the microphone test.
Open a blank Word document or EHR test field. Place the cursor, hold Ctrl + Windows, speak one short sentence, then release. Review the inserted text.
macOS
Mac needs microphone access for dictation and Accessibility permission to place text into the app where you are working.
Open the signed DMG, drag DictaFlow Medical into Applications, then launch it from Applications.
Use Apple, Google, or Microsoft with the exact approved email. If Apple offers “Hide My Email,” choose Share My Email for a pre-authorized clinic account.
In System Settings → Privacy & Security, enable Microphone and Accessibility for DictaFlow Medical. Then open DictaFlow Settings and set Domain to Medical & Clinical.
Open a blank document or test field. Place the cursor, hold Fn, speak one short sentence, then release. Review the inserted text.
iPhone and iPad
The native Medical app guides you through recording, review, and reuse on iPhone and iPad.
Download the Medical app from the App Store. Confirm the app name includes “Medical.”
Use the same approved email as your desktop account. If you use Apple sign-in for a clinic pilot, choose Share My Email.
Tap Allow when iOS asks for microphone access. DictaFlow cannot record until this permission is enabled.
Start a new dictation in the app, speak naturally, then stop. Review the result before copying, sharing, or inserting it into your clinical workflow.
Android
Use the floating microphone with your current keyboard, enable the DictaFlow Medical keyboard, or set up both.
Open the Google Play listing and install the Medical app.
Use Apple, Google, or Microsoft with the exact approved email, then review and accept the Medical data-processing disclosure.
Floating microphone keeps Gboard, Samsung Keyboard, or SwiftKey. DictaFlow keyboard gives you a dedicated voice keyboard. Quick Start can configure either or both.
All methods need Microphone. Floating Dictation also needs Display over other apps and Accessibility; the dedicated keyboard must be enabled in Android keyboard settings. Finish with a short practice dictation.
Your first dictation
Use a blank document or non-PHI test field first. Once the result lands correctly, repeat the same workflow in your approved clinical environment.
DictaFlow produces a draft. Check names, medications, measurements, negations, dates, and clinical meaning before saving, signing, billing, or acting on it.
Click exactly where the final text should appear. DictaFlow returns to that active field after processing.
Hold the trigger on desktop or use the record control on mobile. Speak in phrases. Natural pauses are fine.
Release the trigger or tap stop. Wait for the result, then check it before moving on.
Desktop output
Windows and Mac can insert the same final text in two different ways. The content does not change—only how it reaches the active field.
DictaFlow places the finished result through the clipboard. This is usually the fastest option for Word, Outlook, browser forms, and ordinary EHR fields.
Choose Type characters in the Output setting. DictaFlow sends the finished result as keystrokes for Citrix, RDP, VDI, VMware Horizon, or fields where ordinary paste is blocked or unreliable.
Clinic rollout checklist
Start with one provider. Run the local and remote tests. Once they pass, add the rest of the clinic in stages.
For vendor review, see the security overview, BAA information, and admin guide.
Plan a clinic rolloutFinish vendor review before PHIComplete the security review and any required BAA before staff use DictaFlow Medical with patient information.
Set up the first providerInstall the Medical app, choose the intended microphone, and confirm the clinician's Medical account is active.
Test local cursor insertionIn an eClinicalWorks test field or another approved EHR, dictate fictional text only. Review it, then discard it.
Test the hospital sessionIn Citrix or another remote desktop, try Paste Mode first. If the clipboard fails, use Type characters and increase Typing Delay if letters are skipped.
Check the microphone and its buttonsIf the computer and DictaFlow show the device as an audio input and the level moves, it can capture speech. A proprietary record button may need vendor software or shortcut mapping.
Use one login per clinicianHave each provider use their own work email and account on approved devices. Never collect passwords or use one shared clinic login.
Expand in stagesAfter the first provider passes both tests, onboard the next group. Check each person's local EHR and remote session before live use.
Arrange clinic billing firstContact DictaFlow Medical sales before staff buy plans. Coordinate clinic onboarding and billing there instead of setting up separate individual subscriptions.
Useful after day one
You do not need to configure everything before your first dictation. Add these as you notice repeat needs.
Add medications, clinician names, facilities, abbreviations, specialty terms, and uncommon spellings. Vocabulary informs transcription and AI cleanup.
Examples“empagliflozin” · “Mordechai Weinreb” · “FNP-BC”
On Windows, save frequently repeated phrases or templates under Text Expansions. Labels may differ on other platforms.
Example“normal neuro exam” → your approved normal-exam template
Use AI Smart Edit on Windows, AI Mode on Mac and Android, or AI Refine on iPhone and iPad to clean up wording or revise selected text. Creating new content from voice requests is an optional advanced setting.
Example“Make this concise and format it as Assessment and Plan.”
Set formatting rules for headings, bullets, brevity, and fact preservation. Use AI Smart Edit when you need the note restructured.
Tested custom instructions
On Windows, open Settings → Dictation Behavior → Prompt → Edit Custom Instructions, paste one prompt, then choose Save & Close. Set Settings → Dictation Behavior → Domain to Medical & Clinical. Under Settings → Shortcuts & Interface → AI Smart Edit, confirm AI Smart Edit is enabled, note or change its shortcut, and choose its Output behavior. The Windows default shortcut is Alt+Windows.
These are formatting aids tested on synthetic examples, not clinical validation or a preservation guarantee. Before charting, compare the result with the source and verify names, medication names, doses, routes, allergies, negations, uncertainty, laterality, dates, results, and whether an action was proposed, ordered, completed, or not done.
Windows workflow: To edit existing text, highlight it, use AI Smart Edit, say the immediate edit request, then stop recording. Selected-text transformations work even when voice-request generation is off. With no selection, place the cursor and dictate through AI Smart Edit so standing Custom Instructions can shape the new text. To create a new artifact from a spoken request, enable Settings → Show advanced options → Advanced Configuration → Generate content from voice requests (AI Smart Edit); it is off by default, and without it requests are generally preserved as text.
Treat what you say during AI Smart Edit as the immediate task and Custom Instructions as standing guidance. Keep them compatible. If they conflict, results can vary; rewrite or temporarily remove the standing rule instead of relying on one instruction to override the other. DictaFlow's safety and mode rules still apply.
Format only as a concise SOAP note with exactly these headings: Subjective, Objective, Assessment, Plan. Use only dictated facts. Preserve every negation, uncertainty, medication, dose, and conditional plan. Add nothing.
Evidence: controlled production-model formatter calls using the exact deployed formatter and prompt builder; two repeats passed. ASR and authenticated production HTTP were bypassed.
Before, syntheticThe patient reports three days of dry cough and fatigue. Denies chest pain and denies shortness of breath... pneumonia cannot be excluded...
Abridged from one actual output; … marks omitted text…
Assessment:
Likely viral upper respiratory infection, but pneumonia cannot be excluded.
Plan:
Supportive care. Chest X-ray only if symptoms worsen.
Format only as a concise consult note with headings Reason, Findings, Impression, Recommendation. Keep uncertainty and proposed actions exactly as dictated. Add no greeting, diagnosis, result, medication, or follow-up.
Evidence: controlled production-model formatter calls using the exact deployed formatter and prompt builder; two repeats passed. ASR and authenticated production HTTP were bypassed.
Scope note: tested on one synthetic palpitations consult without medication doses. It is not evidence that doses will be preserved in other dictation. Use it only for the matching task and complete the clinical review above.
Before, syntheticEpisodic palpitations. ECG showed sinus rhythm at 82 beats per minute... No troponin result was provided...
Abridged from one actual output; … marks omitted text…
Impression: cause remains uncertain.
Recommendation: a 14-day ambulatory monitor may be arranged; it has not been ordered.
Format only; do not omit or summarize any sentence. Use headings Medications, Allergy/Correction, Assessment, Plan, and Other/Uncertainty when needed. Preserve adherence uncertainty, do-not-label statements, full correction history, and exact medication names, doses, routes, and frequencies. Add nothing.
Evidence: controlled production-model formatter calls using the exact deployed formatter and prompt builder; two repeats passed. ASR and authenticated production HTTP were bypassed.
Why Other/Uncertainty mattersA shorter broad prompt dropped “adherence is uncertain” in testing. The explicit Other/Uncertainty section preserved it in both repeats.
Abridged from one actual output; Assessment and Plan omitted hereAllergy/Correction: patient first said no medication allergies, then corrected this to rash with amoxicillin.
…
Other/Uncertainty: adherence is uncertain; do not call it noncompliance.
Format only; do not rewrite clinical words. Keep PO, BID, and PRN uppercase exactly. Put each dictated plan item on its own bullet. Preserve all doses, timing, negations, and uncertainty. Add nothing.
Evidence: controlled production-model formatter calls using the exact deployed formatter and prompt builder; two repeats passed. ASR and authenticated production HTTP were bypassed.
Before, syntheticContinue propranolol 40 milligrams PO BID. Sumatriptan 50 milligrams PO PRN at onset...
Abridged from one actual output; … marks omitted text…
Plan:
- continue propranolol 40 milligrams PO BID.
- sumatriptan 50 milligrams PO PRN at onset; may repeat once after two hours.
- return in six weeks.
…
Preserve every conflicting statement. For actual conflicts only, add exactly one final sentence beginning 'Unresolved conflict:'. Do not label ordinary uncertainty, absent exam details, or an uncompleted action as a conflict. Do not choose a side. Keep proposed actions proposed and uncompleted. Add nothing else.
Evidence: controlled production-model formatter calls using the exact deployed formatter and prompt builder; two repeats passed. ASR and authenticated production HTTP were bypassed.
Scope note: tested on one synthetic laterality conflict without medication doses. The prompt does not positively require preservation of ordinary uncertainty or doses in other dictation. Use it only for a matching contradiction task and complete the clinical review above.
Before, syntheticThe patient initially said right knee, then corrected to left knee. Later... the right knee felt swollen. MRI may be considered... MRI was not ordered.
Abridged from one actual output; … marks omitted text…
Unresolved conflict: the patient reported both left knee pain and right knee swelling.
Start with one output shape, then add preservation rules and a no-invention rule. Avoid stacking several templates or mixing “summarize” with “do not omit any sentence.” A newly combined prompt has not been validated merely because its individual sentences appeared in other prompts.
What testing covered: three generated-audio scenarios passed through the exact deployed transcription dispatcher and, where applicable, the formatter; they were not authenticated production HTTP endpoint tests. The copyable prompts above were controlled production-model formatter calls using synthetic text. This was bounded product testing, not clinician validation or a safety study. Clean generated audio does not cover accents, background noise, multiple speakers, or every platform and model combination.
Mode boundary: In these bounded synthetic tests, AI Smart Edit applied requested structure more often than Command Suite alone, but broad prompts still failed and model output can vary. Review every result against the source. Use Command Suite for stronger transcription guidance, not as a guarantee that structural transformations will be applied.
Five-minute readiness check
Complete these checks once on each device you plan to use.
Get the Medical appCorrect appThe installed app is DictaFlow Medical.
Correct identityYou signed in with the exact approved email.
Permissions readyMicrophone works; desktop input permissions are enabled.
Practice passedA short non-PHI sentence reached the intended field.
Review habit setYou checked the result before saving or signing.
Quick fixes
Check microphone permission, select the intended input device, confirm another app is not holding the microphone, and run the in-app microphone test where available.
Keep the destination field focused until insertion finishes. On Mac, recheck Accessibility permission. On desktop remote systems, switch from Paste Mode to Typing Mode.
Increase Typing Delay in advanced settings, then test one short sentence. Remote desktops may need a slower delay than local apps.
Add the preferred spelling to Knowledge Base. Use the exact capitalization and punctuation you want DictaFlow to preserve.
Confirm you used the exact approved email and the Medical app. Do not use Apple Hide My Email. If the address is correct, contact support with your platform and app version—never send PHI in ordinary email.
Test Paste Mode and Typing Mode with non-PHI sample text. Send support the platform, app version, EHR/remote environment, and a redacted description of what happened.
Keep this guide handy
Install the app, sign in, and try one short non-PHI dictation. Bring any workflow questions to onboarding so the session can focus on your real setup.