Nabla and Epic: Toolbox, Hyperdrive Launch and the Two to Three Week Claim

Three layers, per the vendor's Epic page. First, authentication: SMART on FHIR login, or SSO through SAML with SCIM provisioning. Second, the launch: a button in Hyperdrive that opens Nabla in context, either embedded in the Epic client or in a separate browser window when the microphone needs to live outside the Epic frame. Third, the write path: after the clinician reviews the draft, the note and any suggested problems, visit diagnoses and vitals are exported to the encounter. The Toolbox build adds a fourth path, Epic's own ambient module in Haiku, which reads the visit diagnoses to shape the note and writes back through Epic's framework.

The two to three week figure is Nabla's estimate of its own work. It assumes your Epic team has already approved the vendor programme client, configured SMART launch, completed security review and mapped templates. At most health systems those steps are the long pole, and they are the ones described in our integration questions guide. Treat the vendor figure as the time from your approvals to their go live, not from contract signature.

Not published: which FHIR resources are read and written at a site, the Epic version floors for the Hyperdrive and Haiku builds, and whether any HL7 v2 feed is used for the schedule. Ask for the technical specification your analyst will receive.

The draft note, structured to your templates and shaped by the visit diagnoses already in Epic. Suggested new problems, visit diagnoses and vitals that the clinician accepts or rejects before export. Patient follow-up instructions generated from the same encounter. In Haiku, ambient recording without leaving the Epic app. Nabla's case study data claims 55 percent of users save at least an hour a day; that is a vendor figure from self reported surveys and you should measure your own.

The structured data export is the feature to look at closely. Writing a problem list entry or a diagnosis is a step beyond writing narrative text, because it changes what downstream systems see, including billing and quality reporting. It is useful, and it is the direction the whole category is heading, but it needs a governance decision and an audit sample. Our governance committee playbook covers how to stage that. For the documentation loop itself, the AI medical scribe use case page lists what to measure in the first ninety days.

• The vendor programme client and the SMART on FHIR launch configuration, raised through your own integration request process.
• Security review and the business associate agreement, with audio retention, EU or US processing location, subprocessors and model training settled in writing. The HIPAA and AI page lists what to insist on.
• For the Haiku build, Epic's ambient security points and the Haiku version floor.
• Template mapping, and a decision on whether structured export of problems, diagnoses and vitals is enabled at go live or later.
• Identity: SAML and SCIM configuration if you are not using SMART login.
• A test plan with named clinicians, and a change control slot.

If you are a smaller group on a hosted Epic instance, the first item may sit with whoever hosts you, and their queue is the one that matters. We size that in an AI readiness audit.

The most common complaint in the pre Toolbox period was export friction: notes had to be copied or exported rather than landing directly, particularly for group accounts. Nabla's Toolbox announcement says explicitly that the new build resolves "previous export limitations". Ask whether your deployment is on the new build, and test the write back with your own templates before signing off.

Second, the browser window option. Running Nabla in a separate browser window beside Epic is a legitimate configuration for audio capture, but it is a second window, and the Epic page explains why second windows reduce adoption. Prefer the embedded launch or the Haiku build wherever your Epic version allows.

Third, price transparency is partial. Nabla publishes an individual clinician plan but quotes organisation pricing on request. Our Nabla pricing page sets out what is known, and the scribe pricing comparison puts it beside the other Epic embedded vendors.

• Does the two to three week estimate start at contract signature or at completion of our Epic approvals, and what does it assume we have already done?
• Which build are we getting, Hyperdrive embedded, browser window or Haiku Toolbox, and what are the Epic version floors for each?
• Exactly what is written back, in what form, and can structured export of problems, diagnoses and vitals be disabled by user class?
• Where is audio processed and stored, in which country, for how long, and is any of it used to train shared models? Put it in the BAA.
• What is the per clinician price at our volume, the minimum term, and the exit terms if adoption stalls?
• Which reference customers run our Epic release, and can we speak to them without the vendor present?

Our vendor HIPAA question list extends the compliance side.

Nabla's Epic story is credible and its timeline claim is more specific than most, which is exactly why it needs checking against your own approval process rather than accepted at face value. An independent review confirms which build your Epic release supports, tests the write back with your templates, and compares note quality and price against Abridge, Dragon Copilot, Ambience and Suki on your clinicians' encounters. We take no vendor commissions and we have no stake in which of the five you pick.

Our vendor selection service runs that comparison and our deployment roadmap sequences the Epic work afterwards. Book a call with your Epic release and your clinician count.