Duplicates found and proposed for merging — never merged.
Data Janitor
Once a week, early Saturday, it reads your customer, supplier, and contact records, clusters likely duplicates by name, email, and tax ID, and drafts a merge proposal listing the evidence for each cluster. Every merge is a decision a person makes in the system.
Like every Infinary agent, it drafts — a person on your team approves before anything is posted.
What it does
The Data Janitor scans master data for the mess that accumulates: 'Acme Inc', 'Acme Corporation', and 'ACME' as three customers; addresses formatted four ways; contacts that are obviously the same person. It clusters duplicate candidates by fuzzy name, matching email, and tax ID, flags malformed addresses, and drafts a merge-proposal report with the evidence per cluster. It never merges, edits, or deletes a record — proposing is the whole job, and records it cannot read are reported as unknown by name.
- Status
- Beta
- Runs
- Saturdays at 1 a.m.
- Cost per run
- Up to $3
- Review
- Creates drafts only — you approve before anything is posted.
- What it reads
- your existing reports, your records, live totals and history
The Problem Today
"Three spellings of the same customer means split order history, wrong credit pictures, and a sales desk that can't find the account. Cleaning it up by hand is the chore that never reaches the top of anyone's list."
How Infinary Handles It
The chore runs weekly on its own, and what lands on a person's desk is a short, evidenced list of decisions — merge these two, fix that address — instead of forty thousand rows.
What a run looks like
What a run looks like
Say a business absorbs two competitors and inherits their customer lists. Saturday's run clusters the obvious collisions — the same tax ID under three names, the same buyer under two emails — and drafts the proposal with the evidence lined up per cluster.
The office manager works the list over the following week, approving the clear merges and skipping the two clusters that turn out to be a father and son with the same name at the same address. Exactly the judgment call software shouldn't make — and this one doesn't.
Data Janitor
How It Works
Step
By Step
Read
It reads your customer, supplier, and contact master records.
Analyze
It clusters likely duplicates by fuzzy name, email, and tax ID, and flags malformed addresses.
Draft
It drafts a merge-proposal report — each cluster with the evidence for why it's probably one record.
You approve
A person reviews each proposal and performs any merge in the system. The agent never does.
What Changes
What You Get
Customers, suppliers, and contacts are re-scanned every week as new records arrive.
Every proposed merge shows the name, email, and tax-ID matches behind it.
It never merges, edits, or deletes — every change to a record is made by a person.
