One copilot.
Inside all four.
Answers from the chart. Reasoning on demand. Action across surfaces.
- Answer 9:21a “Last A1c?”
- 9:21a 8.1% · 11 months ago. Flagged at chart prep.
- Execute 9:21a “Order a retake.” Queued to Script · Pended 9:41a
You ask.
Amelia works.
Amelia doesn't live in a new window, a new app, or a new queue. Amelia runs inside Scribe, Coder, Script, and Inbox: whichever surface the clinician is already working in. Three things Amelia does, each grounded in the same chart the four surfaces built.
Ask the chart.
Get the chart.
Last TSH and the date. Current statin and when it was started. The name of the specialist who saw them for the knee. Pulled from the discrete data the other surfaces have written. Every answer carries a citation the clinician can open.
In Scribe, mid-visit: thyroid follow-up. Clinician asks "When was the last TSH?" Amelia returns the value and the date, source lab linked, rendered inline in the Scribe panel. The clinician doesn't leave the note to get the answer.
Stitch the context.
Return the reasoning.
The chart is rarely one number. A question about a med is also a question about the history that put it on the list, the labs that justified it, the complaints since, and the guideline that shaped the dose. Amelia stitches context across Scribe's notes, Coder's problem list, Script's order history, and Inbox's results. Differential support. Interaction questions, answered. Care-gap surfacing. Guideline lookup, grounded in the patient's specific chart.
In Script, mid-visit: new elevated BP. Clinician asks "What's on their med list that could be contributing?" Amelia reviews the active meds, cross-references interaction patterns with the lab panel Inbox triaged, and surfaces two candidates with rationale. The clinician reads two lines and decides.
One instruction.
Drafts where
they belong.
Order a lab: it's a draft in Script's signing queue. Add a diagnosis to the problem list: it's a draft in Coder's queue. Close a visit with a specific plan: it's the close in Scribe's note. Amelia has no independent write path. Every execution goes through the same discrete-data writes the surfaces use, and still waits on the clinician's sign-off.
End of visit: clinician tells Amelia "Order a follow-up lipid panel in three months and close the note with the plan." Amelia drafts the order in Script with the protocol-match flag and extends the plan line in the Scribe note. Two surfaces, one instruction, one place for the clinician to sign.
Nothing transmits without you.
Amelia drafts; the clinician signs. Every order waits in Script's signing queue until sign-off.
actions transmitted without a signature.
By design, on every CarePilot surface
Where Amelia
is grounded.
Amelia doesn't have its own data layer. Amelia uses the chart the other four surfaces built.
Answers come from the discrete data the surfaces have already written: the structured note from Scribe, the ICD-10 codes and problem list from Coder, the active medication list and order history from Script, the lab results and renewal history from Inbox. When Amelia returns a value, the source is linked. When Amelia returns reasoning, the chart facts that grounded it are citable. There is no separate index or synthetic layer for the clinician to cross-reference against the original.
Amelia doesn't create a second order queue or a shadow inbox. When Amelia drafts an order, it's Script's draft in Script's signing queue. When Amelia extends the plan in the note, it's Scribe's queue. When Amelia adds to the problem list, it's Coder's queue. The discrete-data architecture Scribe, Coder, Script, and Inbox are built on is the architecture Amelia lives inside.
Every draft and plan extension is a clinician-reviewable artifact. The assistant is an assistant. The clinician signs. The judgment stays.
See Amelia run
in your practice.
30 days free. Then you decide.