Multi-EHR and locum tenens AI scribe 2026: one workflow, many charts

May 8, 2026 · 6 min read

Native ambient AI scribes are EHR-locked. Suki on Athena, Sunoh on eCW, Oracle Voice on Cerner, Heidi as a partner across systems — each requires the clinician to be inside the host EHR's workflow. That's fine for the 70% of clinicians employed by a single system. It's broken for the locum tenens, multi-system employed, hospitalist consulting, and surgery-center mix groups who work across 2-5 EHRs in a typical month.

For multi-EHR clinicians, the only stack that actually works in 2026 is the EHR-agnostic DIY pipeline: external transcription, clinician-controlled prompt, structured output, copy-paste or generic-FHIR write into whichever chart is in front of you that day.

The multi-EHR clinician profile

Why native ambient scribes don't fit

• Authentication friction. Each native scribe has its own auth, MFA, and session timeout. Switching EHRs means re-authenticating, often re-pairing the recording device.
• Per-system pricing. Each EHR's native ambient charges separately. Five EHRs at $200-300/provider/month = $1,000-1,500/month for one clinician.
• Inconsistent prompt / template. Each native scribe outputs in its own template style. The clinician's note quality varies by which chart they're in.
• BAA chain explosion. Each EHR + each native scribe = separate BAA. Five EHRs = five BAA chains to manage.
• Specialty fit varies. Native scribes tuned for primary care don't handle the locum-hospitalist's procedural notes well, etc.

The multi-EHR DIY pipeline

1. Audio capture (system-agnostic). Phone-based recording app, dedicated handheld recorder, or laptop-mic during the visit. Same hardware, every chart, every day.
2. Transcription (single vendor). LessRec at $0.05/min — works regardless of which EHR you'll write into. No subscription floor.
3. LLM with clinician's preferred prompt. One prompt the clinician owns. Same SOAP / structured note style, every visit, regardless of EHR.
4. Structured output. Markdown or plain text that's easy to paste into any chart's note field.
5. Per-EHR formatting tweaks. Optionally, second pass formats the output for the specific EHR's quirks (Epic note tags, Cerner section dividers, etc.).
6. Copy-paste into the active chart. 30-60 seconds per encounter, across any EHR.

The math: one clinician across five EHRs

For a locum tenens working 18 days/month at average 12 patients/day = 216 encounters/month. Audio at 18 min average = 3,888 minutes = $194/month at LessRec rates — but more realistically the clinician records selectively, so $50-100 covers actual volume.

The clinician-owned prompt

The single largest advantage of the multi-EHR DIY stack is that the clinician owns the prompt. One prompt, refined over months, produces consistent SOAP notes across every EHR. The native scribes can't replicate this — their prompts are vendor-controlled and EHR-specific.

You are documenting a clinical encounter for a multi-EHR clinician.

INPUT:
- Encounter audio transcript (verbatim)
- Visit context provided by clinician: patient ID, age, chief complaint, EHR target

OUTPUT a structured note in this format (Markdown):

## Subjective
[HPI from patient + clinician transcript, structured by symptom + duration + modifiers]

## Objective
[Vitals if mentioned, exam findings cited from transcript]

## Assessment
[Each problem on its own line, with ICD-10 at maximum specificity supported by the transcript]

## Plan
[Per-problem plan, including meds, labs, imaging, referrals, follow-up timing]

## Patient instructions
[What the patient was told to do, in patient-facing language]

For each clinical fact, cite the transcript line that supports it. Flag any decision-relevant content that's ambiguous.

This prompt produces a note that pastes cleanly into Epic, Athena, eCW, NextGen, Cerner, EMA, or any other EHR with a free-text note field.

Specialty considerations for multi-EHR clinicians

Locum hospitalists: the prompt should adapt for admission notes, progress notes, and discharge summaries. Three prompt variants, one workflow.

Multi-EHR specialists (cardiology consulting across hospitals, ortho across surgery centers): use the specialty primer + structured op-note prompt from our specialty guides.

Behavioral health across cloud EMRs: DAP/SOAP variant from our behavioral health guide.

BAA chain

The multi-EHR DIY stack needs:

• Audio capture: clinician-owned device or app with documented data flow
• Transcription vendor BAA (LessRec offers BAA on request)
• LLM vendor BAA (Anthropic Enterprise or OpenAI Enterprise)
• For each employer or contracting facility: documented disclosure that the clinician uses this AI scribe stack, signed off by the facility's compliance officer or per the locum staffing agreement

The locum staffing agencies (Weatherby, CompHealth, Locum Leaders, etc.) are increasingly aware of AI scribe use and will document compliance per assignment.

Audit retention

For multi-EHR clinicians, retention policy gets complicated because audio + transcript pertain to records owned by different facilities. Best practice:

• Retain audio + transcript for the longer of: 7 years, applicable state medical board rules, or facility-specific retention requirement
• Tag each recording with the patient ID + facility + date so it's discoverable in any future request
• Deletion on schedule unless facility-specific request to retain longer

When to start

If you work across 2+ EHRs in a typical month, the multi-EHR DIY stack pays off in the first week. The setup is hours: an account, a prompt, a recording app. The reward is one consistent documentation workflow regardless of which chart is in front of you.