Healthcare & Life Sciences
Document AI
Alderleaf Health Network
Clinical intake that clears itself
before staff sit down.
Straight-through intake
14 weeks
91%
Was
0%
+91 pts
4,200 h
Staff hours saved / month
14
Clinics onboarded
14 wks
Kickoff to production
02 · The engagement
Published
01 · Challenge
What was actually broken
Not the symptom, the cause.
Alderleaf’s intake team was manually keying data from referral letters, insurance forms and prior-provider records into its EHR before every new-patient appointment — a process prone to transcription errors and a major source of same-day scheduling delays across its 14 clinics.
02 · Approach
How we scoped it
And what we deliberately chose not to build first.
We audited a sample of 1,200 real intake documents to map the actual variation in layout and quality, not just the clean templates. The extraction pipeline was tuned per field rather than per document type, so a clearly legible patient name doesn’t get held up by a smudged fax header elsewhere on the page. Low-confidence fields route to a review queue pre-filled with the pipeline’s best guess, so reviewers correct rather than retype.
03 · Architecture
The system, in one diagram
What it connects to, what a person still reviews, where the guardrails sit.
Solution architecture
Documents
Referrals, labs, forms
Fax gateway + portal
Document type, provider
Classification
Validated vs patient record
Field extraction
Confidence scoring
Per-field threshold
Pass · Clinical sign-off
EHR staging tables
Else · Low-confidence fields
Intake review queue
Field accuracy by source · alert when a provider format degrades
An extraction and classification pipeline ingests documents from Alderleaf’s fax gateway and patient portal, applies field-level confidence scoring, and writes directly into the EHR’s staging tables for clinical review before finalization. A monitoring layer tracks field-level accuracy by document source and alerts the operations team when a new referring provider’s format degrades performance.
04 · Result
What changed
Measured in production, not in the pilot.
Fourteen weeks after kickoff, the pipeline reached a 91% straight-through processing rate across all intake document types, with the remaining 9% routed to a review queue that takes reviewers roughly a third of the time manual entry used to take.
Result at a glance
Client
Alderleaf Health Network
Industry
Healthcare & Life Sciences
Service
Document AI
Time to production
14 weeks
Headline
91%
straight-through processing rate on intake documents
03 · What changed
In the client’s own operating numbers
91%
straight-through processing rate on intake documents
4,200 hrs
staff hours saved per month across 14 clinics
14 weeks
from kickoff to production
+91 pts
Straight-through intake
0%
→
91%
04 · In their words
Operators, not buyers
Our intake staff used to dread the referral backlog after a holiday weekend. Now the system has already done the first pass by the time they sit down, and they're reviewing instead of retyping. That's hours back in a clinic day that used to disappear into paperwork.
Dr. Aaron Mbeki, Chief Medical Information Officer, Alderleaf Health Network
Next case study
Let’s find out
what it’s worth in production.
Have a workflow like this one? A 30-minute call is enough to tell whether it is ready to build.