Workflow automation · identity resolution

intake-router-multisource

Routing fast is useless if the same person arrives through different channels and gets treated as separate pipeline.

Public · synthetic demo

The problem

Duplicate identities create duplicate work and misleading pipeline.

A form submission, an ad lead, an inbound email, and a CSV row do not share a schema. Routing each record independently can double-contact one person or silently merge two different people who happen to share a name.

Input

Heterogeneous records enter one canonical decision layer.

Sources ingested:
csv_import
email_inbound
facebook_lead_ads
web_form

Raw records: 17

The money shot

Strict matching reduces 17 raw records to 15 identities without guessing.

Raw records

Canonical leads

Auto-merge events

Ambiguous pairs not merged

Human-review boundary

Five review reasons cover three distinct leads. Exact email merged Sarah Chen at confidence 0.99. Exact phone merged Marcus Webb at 0.90. Two David Kim records stayed separate because both contact methods differed.

How it's verified

Every merge and route leaves evidence.

Output	Measured result
`merges.csv`	2 rows with exact rule and confidence
`review_queue.csv`	5 reasons across 3 leads
`routing.csv`	15 leads with queue, owner, SLA, priority, and region

Honest limitations

This resolves a fixture, not every real identity.

The data is synthetic and the normalization vocabulary is illustrative.
Name and company similarity only trigger review. They never prove identity.
The router stores no lifecycle state and is not a CRM.
The optional model path summarizes messages; it does not control identity resolution or routing.

Every automatic merge carries provenance; every uncertain one stops for a person. jigonyoo.com · Back to hub