AI Intake

A forty-page fax becomes a patient record.

Referrals arrive as scans, merged packets, and pages out of order. Bosshh reads them into a structured record and tells your coordinator the one thing they need to know first: what is blocking this admission.

Admission decisions turn on a short list

Most of a referral packet is not what determines whether you can admit. Bosshh establishes each gating item explicitly — including when it is absent, which is the finding that actually matters.

Physician order and signature

Whether a valid signed order is present, and the signature date.

Face-to-face encounter

Whether it is documented, and when it happened.

Certifying physician and NPI

Who is certifying, with the identifier you need for billing.

Coverage and authorization

Payer, plan type, member ID, and whether an authorization number exists.

Homebound justification

The stated reason, in the referral's own words.

Ordered disciplines and frequency

What was ordered, at what visit frequency, per discipline.

Built for messy documents

Conflicts are reported, not resolved silently

When page three and page eleven disagree about a date of birth, Bosshh gives you both readings and flags the conflict. Picking one quietly is how a wrong DOB ends up on a claim.

Ambiguity stays ambiguous

An illegible date is left empty and noted, not guessed into ISO format. Names, doses, and identifiers are extracted as written rather than tidied up.

Gaps come with an action

Each gap says what is missing, how badly it blocks the admission, and what the coordinator should do about it — call the referral source, request the order, verify the ID.

Eligibility checked alongside

Coverage data is validated as part of the same pass, so the coordinator sees the extraction and the coverage picture on one screen. Plan types that usually need prior authorization are called out before the first visit is scheduled.

Send us a referral packet

A real one, with its real problems. You will see exactly what the extraction catches.