AI Contact Cleanup
Duplicates pile up with every phone switch and CRM import. This tool fixes the mechanical problems — capitalization, name fields, phone and address formats — in your browser, then uses AI to catch the duplicates exact-match tools always miss.
What Gets Fixed
🔤 Capitalization
- • JOHN SMITH and john smith both become John Smith
- • McDonald, O'Brien and MacLeod keep their internal capitals
- • Jean-Luc and other hyphenated names capitalize both halves
- • van der Berg keeps its lowercase particles
- • Names already in mixed case are left alone
✂️ Name Splitting
- • A single name field becomes first / middle / last
- • "Smith, John A." is reordered correctly
- • Dr., Prof. and other honorifics move to the prefix field
- • Jr., III and PhD move to the suffix field
- • Multi-word surnames like "van der Berg" stay together
📱 Phone Normalization
- • Mixed formats collapse to one consistent style
- • 00-prefixed international numbers become + numbers
- • Local numbers gain a country code you choose
- • Extensions are preserved, not thrown away
- • Vanity and SIP numbers are left untouched
📍 Addresses & Emails
- • Street and city names get consistent capitalization
- • US state codes are uppercased; longer names title-cased
- • Postal codes are uppercased and trimmed
- • Email addresses are lowercased
- • Bare domains become full https:// URLs
Two Kinds of Duplicate Detection
Groups contacts that share an unambiguous identifier. Instant, free, and nothing leaves your device.
- • The same email address
- • The same phone number, however it is formatted
- • An identical name after accents and punctuation are folded
- • Chains: A matches B, B matches C, so all three group together
Catches the duplicates that no exact-match rule can, by reasoning about how the same person gets written down differently.
- • Nicknames: Jon/Jonathan, Bill/William, Kate/Katherine
- • Company variants: "Acme Corp" = "ACME Inc."
- • Maiden and married surnames for the same person
- • Initials against full names, and transliteration differences
- • Each match comes with a confidence score and its evidence
Merging Never Loses Data
A bad merge destroys information you cannot get back, so merging is deliberately conservative:
- The most complete entry wins. Whichever contact has the most filled-in fields becomes the base record.
- Nothing multi-valued is discarded. Every phone number, email address, postal address and URL from every entry in the group is kept.
- Empty fields are filled from the others. If the base record has no job title and a duplicate does, the title carries over.
- Notes are concatenated, not overwritten.
- You approve every merge. Each group shows a preview of the merged result and a "keep separate" button. Nothing is applied until you download.
What Leaves Your Browser
Parsing, normalization, exact-match detection, merging and export all run locally — your contact file is never uploaded. The one exception is AI fuzzy matching, and even then only a reduced summary is sent.
Sent for AI matching
- • Name, company and job title
- • Email domains only (example.com)
- • The last four digits of each phone number
- • City
Never sent
- • Full email addresses
- • Complete phone numbers
- • Street addresses
- • Notes, birthdays and photos
Free personal email domains such as gmail.com are dropped too, since they say nothing about whether two people are the same. If you would rather send nothing at all, the exact-match pass alone still catches shared emails and phone numbers.
Perfect For
📱 After Switching Phones
Restoring a backup on top of a synced account is the classic way to end up with two of everyone. Merge them back into one list.
🔗 After a CRM Import
Imports rarely match on anything but exact email. Fuzzy matching finds the records the importer created twice.
🗃️ Merging Multiple Accounts
Combining work, personal and old accounts produces overlapping entries with slightly different spellings of the same people.
📇 Before an Email Campaign
Consistent names and one entry per person means no one gets the same message twice addressed two different ways.
Frequently Asked Questions
How is this different from the free duplicate detector?
The duplicate detector scores contacts against each other with fixed rules and asks you to delete the extras. This tool also standardizes every field, merges groups into one complete record instead of deleting, and adds an AI pass that recognizes nicknames and company-name variants that scoring rules cannot express.
Can the AI merge two different people by mistake?
It is instructed to skip uncertain matches rather than risk it, and to treat a shared employer or email domain as supporting evidence only — never as proof on its own. Matches below 55% confidence are discarded outright. Every remaining group is shown with its confidence and reasoning, and you can mark any group "keep separate" before downloading.
Does it work with contacts in other languages?
Yes. Scripts without letter case — Japanese, Chinese, Korean, Arabic, Hebrew, Thai — are left untouched by the capitalization pass rather than mangled, and accented Latin names are compared with accents folded so "José" and "Jose" match.
What is free and what needs premium?
All field standardization and exact-match duplicate detection are free, within the site's monthly contact allowance. AI fuzzy matching requires premium, as does processing lists above the free contact limit.
Will my vCard formatting survive?
Output is regenerated with correct CRLF line endings, escaped values and folded long lines, and you choose the vCard version. Files with QUOTED-PRINTABLE encoding or Apple-style grouped properties are read correctly on the way in.
Can I undo a merge?
Nothing is changed until you download. Every group has a "keep separate" toggle, and your original file on disk is never modified.
Related Tools
Photograph business cards and send the results straight into this cleanup tool.
Combine several VCF files into one before cleaning them up together.
The simpler, rule-based tool for removing exact duplicates.