← Back to projects
04 / Legal / AI
Referral Intelligence Platform
Referrals were coming through scanned PDFs, email attachments and intake forms. The intake process depended on staff retyping the same information and checking for missing fields. We built one pipeline that reads, validates and routes every referral.
86%reduction in average intake time
1.3 minmedian processing time per referral
7.2 hrsaverage weekly admin time recovered
What we built
A system designed around the actual work.
- Automated email and document capture
- OCR for scanned referrals
- Client and case detail extraction
- Missing‑field detection
- Duplicate referral checks
- Structured case creation
- Trackable intake queue
How it works
From input to action through one connected flow.
01
RECEIVE
Capture fax, email or form submission
02
READ
Run OCR and extract required fields
03
CHECK
Detect missing or duplicate information
04
TRACK
Create and route the intake record
Delivery scopeSolution architecture, workflow design, AI integration, application engineering and production implementation.
TechnologyOCR · n8n · LLM extraction · ClickUp
Measured outcome
Operational work became visible, measurable and easier to scale.
Average intake time moved from 9.6 minutes to 1.3 minutes for referrals that passed confidence checks. Exceptions were visible immediately instead of surfacing later in the process.