Ambient Clinical Documentation
How to instruct Copilot during and after patient encounters for accurate ambient notes.
Complete training guide for clinicians: from ambient documentation to "Prepare My Note" automation. Open in any browser, fully interactive.
Dragon Medical Copilot uses generative AI: not transcription. How a doctor phrases a prompt fundamentally changes the output. This training teaches clinicians to write prompts that are specific, contextual, and format-aware.
How to instruct Copilot during and after patient encounters for accurate ambient notes.
Designing specialty-specific note prompts that produce consistent, structured output.
Voice commands for editing, formatting, inserting, and correcting within a draft note.
Prompts for patient summaries, referral letters, portal messages, and discharge notes.
ICD-10, CPT, eCW, billing, and safety flag prompts that run automatically after every note.
Unlike older Dragon versions that transcribed every word, Copilot interprets and structures content. What the doctor says, and how, determines what gets written.
Instructions like "Document this as a SOAP note for cardiology" before speaking will change the output structure entirely.
Prompts that include format, tone, audience, and required sections dramatically cut post-generation review time.
Custom instructions saved in templates allow Copilot to "remember" how each physician likes their notes formatted.
Post-note prompts execute after every note: adding codes, flagging safety issues, and formatting for the EHR without any extra physician action.
Ambient AI captures the natural conversation between doctor and patient and converts it into a structured clinical note. The prompt spoken before or after the encounter frames how Copilot interprets and documents that conversation.
More elements = more precise output. Even 2-3 elements significantly improves the note.
A primary care physician about to see a routine visit patient:
Why it works: States note type (follow-up), specialty (primary care), and format (SOAP). Copilot organizes Subjective, Objective, Assessment, and Plan automatically.
A cardiologist seeing a patient for chest pain evaluation:
Why it works: Specialty-specific sections are named explicitly. The flag instruction prompts Copilot to highlight drug categories, saving a review step later.
A psychiatrist conducting a 60-minute intake session:
Why it works: Adds a critical safeguard, mandates clinical language, and builds in a safety flag supporting risk documentation workflows.
"Document my visit."
No structure guidance. Output will be a flat paragraph, harder to review and less EHR-ready.
"Document this orthopedic follow-up as a progress note. Include pain score, ROM findings, imaging review, and updated treatment plan."
Produces a structured, section-labeled note that slots directly into the EHR.
Specialty-specific templates let clinicians set recurring preferences once, then reuse them for consistent structured notes.
Why it works: Set once, applies to every visit. The physician never has to re-explain preferences.
Why it works: The closing recommendation paragraph for the referring physician eliminates a common documentation friction point in cardiology.
Why it works: ROM measurements, pain scores, and weight-bearing status are orthopedic-specific data points that Copilot will specifically capture from the conversation.
Why it works: The explicit safety documentation protects the physician medico-legally. Coding complexity tagging supports billing accuracy.
Why it works: Percentile flagging and anticipatory guidance bullets are pediatrics-specific: this template would take 25+ minutes to write manually per patient.
Why it works: Disease activity scores and biologic dates are recurring rheumatology documentation requirements. Naming them explicitly ensures Copilot captures them from every conversation.
Once a note draft is generated, clinicians can edit, reformat, add, and correct using natural language voice commands without touching a keyboard.
| Action | Command Example | Tips |
|---|---|---|
| Add a section | "Add a section called Allergies after the medications section" | Name the new section and anchor it to an existing one. |
| Delete content | "Remove the sentence about ibuprofen in the Plan" | Specify which section to avoid deleting the wrong content. |
| Condense a section | "Condense the Assessment into one concise paragraph with the primary diagnosis first" | Name the target section and the exact length or structure. |
| Replace terminology | "Replace hypertension with HTN throughout the note" | Use global replace language when the change should apply everywhere. |
| Reformat | "Convert the Plan section into a numbered list of action items" | State the destination section and the desired format. |
| Change tone | "Rewrite the patient instructions at a 6th grade reading level with no abbreviations" | Always specify the audience and reading level. |
| Move content | "Move the imaging review sentence from HPI to Objective" | Identify the source and destination sections. |
| Add a safety item | "Add a medication safety note at the top if any interactions are mentioned" | Make safety flags visible and labeled. |
| Create follow-up | "Draft a portal message under 150 words summarizing the lab results" | Add length, audience, and purpose. |
| Finalize | "Add Billing Documentation at the end with MDM level and suggested E/M code" | Use a consistent label so reviewers can find it quickly. |
Beyond the visit note, Dragon Copilot can generate patient summaries, referral letters, after-visit instructions, and follow-up messages from a single prompt.
After completing a diabetes management visit:
Why it works: Reading level instruction and action-item format make this directly usable for the patient portal without physician editing.
Referring a patient to a gastroenterologist:
Why it works: The "clinical question" instruction ensures the referral has a clear ask, which most specialists require for triage.
Explaining lab results through the patient portal:
Why it works: A word limit, audience, and action step make the message usable without rewriting.
Preparing follow-up instructions after discharge:
Why it works: It separates clinical documentation from patient-facing instructions and gives the patient clear next steps.
Prepare My Note prompts run automatically after a note is generated. They can add codes, flag safety issues, format for the EHR, and prepare billing support before physician review.
Conversation becomes the source material
Copilot creates the clinical note
"Prepare My Note" instructions execute automatically
Physician reviews fully prepared note
Why it works: The silent omission instruction is critical. Without it, Copilot writes "(No code found)" next to uncertain diagnoses, cluttering the note and creating documentation liability.
Why it works: Same silent omission logic for procedures. It keeps the note clean even when exotic or unlisted procedures are discussed.
Why it works: Specificity prompts encourage laterality, acuity, and episode detail instead of generic coding suggestions.
Why it works: eCW-specific formatting reduces cleanup before the note moves into the chart.
Why it works: The referral block gives staff a clean handoff with specialty, reason, urgency, and codes in one place.
Why it works: The primary diagnosis label supports chart review and billing workflows without adding clutter.
Why it works: This auto-captures the breakdown when the physician mentions total time or MDM elements.
Why it works: These are common sources of first-pass claim denials. Auto-checking catches issues before they reach the billing team.
Why it works: The "omit if clear" instruction trains physicians to treat the section's presence as the alert. A clean note means no issues were flagged.
Why it works: The placeholder creates a mandatory documentation reminder. Medico-legally, an unsigned safety assessment placeholder is better than a missing one.
Why it works: Critical value documentation depends on showing the abnormal value and physician response. The prompt makes that review visible before signing.
Use these as fast reminders during clinician training sessions.
Note type + Specialty + Sections + Tone = precise output
Saved instructions create consistent notes across visits
Specific section names reduce accidental edits
Patient, specialist, and portal prompts need different tone
Tell Copilot what to leave out when no result exists
Scenario exercises reinforce prompt-writing habits
| Prompt Skill | Weak | Better |
|---|---|---|
| State the role | Write my note | "Document this as a cardiology new patient consultation" |
| Set the length | Be concise | "Limit the summary to 3 sentences" |
| List required sections | Include everything important | "Include: HPI, exam, assessment, plan" |
| Add flags | Note anything unusual | "Flag any new medications or abnormal values" |
| Specify tone | Sound professional | "Formal clinical language, third person" |
| Anchor edits | Fix that part | "In the Plan section, change the follow-up interval to 4 weeks" |
| Silent omission | Add codes if you find them | "If no code is found, omit the parentheses entirely" |
Scenario-based exercises for clinician training sessions. Write a prompt for each scenario, then check your answer.
Scenario: Dr. Chen is an internist seeing a 64-year-old for a follow-up on hypertension and type 2 diabetes. She wants a note with vitals, medication review, labs discussed, assessment, and a numbered plan.
Your task: Write the ambient documentation prompt she should say before the encounter starts.
Scenario: Dr. Okonkwo is a rheumatologist who sees patients for RA, lupus, and gout. She wants a saved template that always includes: joint involvement, disease activity score, current biologics with dates, labs (ESR, CRP, ANA), and a flare assessment.
Your task: Write the custom instruction template she should save in Copilot.
Scenario: Copilot generated a 5-paragraph Assessment section. Dr. Patel needs it condensed to one concise paragraph, with the primary diagnosis listed first, and he wants "hypertension" changed to "HTN" throughout the note.
Your task: Write the in-note voice command to accomplish all three edits.
Scenario: Dr. Williams wants to configure a "Prepare My Note" prompt for her primary care practice using eClinicalWorks. She wants ICD-10 codes added silently, a referral block if referrals were mentioned, and a medication safety flag if any issues are found.
Your task: Write the complete "Prepare My Note" prompt for her eCW setup.
Scenario: Dr. Kim is a hospitalist. She needs to: (1) add a discharge summary section, (2) flag medications changed during admission, (3) generate patient-friendly discharge instructions, and (4) draft a follow-up letter to the patient's PCP.
Your task: Write a sequence of prompts to accomplish all four tasks.
Contact Doug Lister directly. A 20-minute call is all it takes to see the difference in a real clinical workflow.
Contact a Microsoft Dragon Copilot ExpertCompare products, review common questions, learn practical Dragon Copilot workflows, and get expert training support for your clinical documentation setup.