The chart is no longer a private letter to your future self
Open notes and the federal information-blocking rules under the 21st Century Cures Act changed the audience for everyday documentation. Many patients can read progress notes soon after signature — sometimes on a phone in the parking lot. That does not mean you should write marketing copy. It means the note must remain clinically precise while surviving a second reader who was in the room and has feelings about how their story was summarized. Accuracy still wins. Casual cruelty, unexplained jargon piles, and speculative asides travel farther than they used to.
Clinicians did not ask for a new audience, but the audience is here. The workable response is craft, not avoidance. Document what you need for continuity, coding, and liability. Skip the venting that used to live safely in a chart almost nobody saw. Teach AI scribes the same manners you teach residents. If your draft tool echoes every tense exchange verbatim, you will spend your evening editing tone instead of confirming plans — which defeats the reason you bought the tool.
What actually changed for clinic operations
Information blocking rules constrain how actors restrict access to electronic health information, with limited exceptions. Practice managers should know who owns portal release settings, how soon notes post after signature, and how sensitive encounter types are handled. Do not assume the EHR’s default matches your specialty’s risk tolerance. Behavioral health, adolescent care, and intimate-partner situations may need pathway-specific workflows developed with counsel and clinical leadership. “Everything posts instantly” and “everything is delayed forever” are both blunt instruments.
Operationally, connect release timing to documentation completion. If notes sit unsigned for days, patients see silence, then a sudden dump of content — and your inbox fills with clarifications. Workflows that produce a reviewed note before the clinician leaves clinic reduce that spike. Whether you use dictation, transcription, or an AI SOAP note draft, the patient-facing clock starts when you sign. Design for that, not only for internal backlog metrics.
Tone without watering down clinical truth
Patients can handle bad news in the chart when it matches what was said in the room. They struggle with surprise pejoratives, unexplained labels, and humor that does not land in text. Prefer observable language over character judgments. “Missed three metformin fills; barriers discussed” beats “noncompliant and difficult.” “Patient declines imaging today; risks of delay reviewed” beats “refuses workup.” You are not writing to flatter. You are writing so the next clinician — and the patient — can see the decision trail.
Keep clinical specificity. Softening “suicidal ideation” into mush helps no one. The craft is pairing precise terms with context the patient heard aloud. If you use a phrase that will alarm on first read, make sure the assessment and plan show what you did next. Open notes punish orphaned scary nouns. They reward complete thoughts. That standard is older than the Cures Act; the portal only made weak notes more visible.
What not to dump into the note
Not every sentence spoken in the exam room belongs in the legal medical record. Side comments about family members, unverified third-party claims, and raw conflict blow-by-blows often create more harm than signal. Document clinically relevant social context carefully. Avoid copying portal messages wholesale when a concise synthesis will do. Do not paste another clinician’s speculative differential into your note as settled fact. AI drafts are especially prone to over-inclusion because models are rewarded for coverage; your review pass should reward discrimination.
Separate psychotherapy notes (where applicable) from progress notes with the discipline HIPAA already expects. Ambient capture makes that separation harder if the microphone is on during spaces that should not become portal text. Define when to pause capture. Teach staff not to discuss unrelated patients in-room while recording is active. The cheapest editing is not generating the wrong content in the first place.
How AI scribes should handle sensitive language
Configure templates and style guides explicitly. Tell the system you want neutral, clinically precise phrasing; prohibit slang; and require that assessment language match problems addressed. Then measure. Pull twenty notes and score them for pejorative phrasing, orphaned sensitive terms, and invented social history. If the model dramatizes conflict, tighten prompts and human review — do not “get used to it.” Our accuracy methodology applies to tone errors as well as drug-name errors because both reach patients.
For specialties with high stigma risk — psychiatry, pain, obesity medicine, substance use — start ambient rollout slower and review more notes per clinician in week one. Consider hybrid paths: ambient for routine follow-ups, dictation for high-volatility visits. Sunrise’s specialty-aware documentation approach exists because a generic model trained on everything writes like it has never sat in your exam room. Demand that awareness from any vendor, and keep a human in the loop for the visits where a clumsy sentence becomes a portal crisis.
Talking with patients when they disagree with the note
Some portal messages will say the note is “wrong.” Sort factual errors from interpretive disagreements. Facts (dose, laterality, test result) get corrected promptly through your addendum process. Interpretations (“you said I was noncompliant”) get a conversation and, if needed, a clarifying addendum that does not erase the clinical record into fiction. Train clinicians not to edit history into quietness after the fact in ways that look like spoliation. Transparency includes knowing how to amend.
Prevent repeat collisions by adjusting the visit itself. Summarize the plan aloud before you stand up. If you document a disagreement, say it respectfully in the room first. Patients who hear the framing rarely ambush the portal. This is also good medicine under time pressure: shared understanding reduces callbacks. Open notes did not invent that habit; they raised the cost of skipping it.
A practical weekly habit for groups adopting open notes culture
Pick five signed notes per specialty each week and read them as if you were the patient with a high-school reading level and a short fuse. Flag three patterns: unexplained acronyms, unsupported labels, and missing plan rationale. Feed those patterns into template tweaks and AI style rules. Share one before/after example in the clinician huddle — not as shame, as craft. Groups that treat note quality as a team sport adapt faster than groups that send a single memo about “being nice in the chart.”
Connect this habit to malpractice hygiene and coding hygiene. Clear plans, documented shared decision-making, and specific assessments protect patients and clinicians. Soft, vague notes fail open notes and fail audits for related reasons. If you want a single documentation north star for 2026, use this: write the note you would defend in clinic, in court, and at the kitchen table with the patient present. Then configure your AI medical scribe and transcription workflows to make that note the path of least resistance.
Where transcription still helps in an open-notes world
Ambient drafts shine in conversational clinic visits. They are weaker when the clinician needs deliberate, composed language — detailed procedure narratives, carefully worded disability opinions, or letters that will be read by schools, employers, and attorneys as well as patients. In those cases, a dictated note reviewed by a trained transcriptionist often produces cleaner first drafts than an ambient capture of a messy room. Open notes raise the value of that craftsmanship rather than eliminating it.
Many groups keep both tools: ambient for routine follow-ups, transcription or structured dictation for high-stakes documents. The patient still reads the result either way. What changes is how much editing you burn to make the tone and structure defensible. If your evenings are spent scrubbing slang from ambient drafts of sensitive visits, that is a routing problem, not a personal failing. Match the capture method to the document’s audience, then keep the same review ritual before signature.