Anatomy of Apple's Personal Knowledge Graph — Eighth Rank · 8rank.ai · El Autómata
Lunes, 16 de noviembre de 2026
Portada de las 22:25
Todos los periódicos y la radio, en unoEscrito por IA a partir de Hacker News · Cómo se hace
Hacker News
8rank.ai · texto original
Anatomy of Apple's Personal Knowledge Graph — Eighth Rank
Every Mac since macOS 13 ships a pair of daemons whose only purpose in life is to build a graph about who you are, whom you know, how those people are related to you and to each other, how you communicate with them, where you go, what you do and when. In this post, we look into how this system works and give you a tool to peek into your own graph.
What if we told you that every Mac ships a pair of daemons whose only purpose in life is to build a graph about you and the people around you: who you are, whom you know, how those people are related to you and to each other, how you communicate with them, where you go, with whom, what you do and when. This graph is built on device by cross-referencing pretty much everything on the machine: Contacts, Calendar, Photos, Messages, Mail, location, Screen Time and a long tail of behavioral signals. And, more importantly, it is held as a queryable store that LLMs retrieve from (allegedly). Would you believe us?
Well, that's what is happening, and interestingly enough, there is little documentation of how this whole mechanism works on your Mac. The closest thing to an official description is the man page (man 8 knowledgeconstructiond), and this is all you get:
man 8 knowledgeconstructiondmacOS 26.6.2, build 25G83
NAME
knowledgeconstructiond – A daemon that analyzes content on the
device to build a general purpose knowledge graph.
DESCRIPTION
knowledgeconstructiond analyzes the content on this device to help
build a knowledge graph representing the device and the user. It
works together with intelligenceplatformd, performing some of the
heavy processing tasks, to generate the knowledge graph.
Previous work: To be fair, Apple's own researchers did publish a paper on the techniques behind this graph construction, back in 2023. But there is next to nothing on how it is integrated into macOS: which versions run it, and the exact list of what is collected and inferred. Also, we personally ran into it in August 2026; but a few months before, Nic Lydon came across the same database and wrote it up in I Built a Personal Knowledge Graph. Apple Had Already Built One on My Laptop. And even more recently, a forensics tool went looking on the phone: in September 2026, Alexis Brignoni taught iLEAPP to read the same store on iOS, where graph.db and the ontology.db that decodes its predicate codes sit under /private/var/mobile/Library/IntelligencePlatform; he writes it up in Reading the iPhone Knowledge Graph, and Checking Its Work. I mean, if forensics folks know about this, we don't see why you should not...
But we go a bit deeper.
This blog post is an account of some analysis work we carried out on what the knowledge graph actually does, so that you (the user) can get a better understanding of how this mechanism works, and some clarity on what the graph holds about you and your family and friends. Everything in it was collected on macOS 26 (26.6.2 at the time of writing).
To this end, we open-source a local dashboard (iKnow) you can use to read the content of your personal knowledge graph on your Mac.
Finally, we derail a bit and look into what we do best: understand how attackers might exploit it. We run a thought experiment (one that applies to any such personalized system) and show what happens if such a knowledge base ends up in the wrong hands, and how a "privacy single point of failure" can be exploited to threaten you in ways you may not have considered before (when combined with an abliterated LLM). We stress that this goes beyond the specific Apple personal knowledge graph, but it generally applies to any such personalized ML / assistant systems (Muse, Dot, etc.)
What's in the personal knowledge graph?
The two daemons keep their work in ~/Library/IntelligencePlatform, a directory of SQLite databases (fifty-one databases) stored in plaintext but located in a zone protected by TCC Full Disk Access.
Likely, the most interesting is graph.db. It is a triple store written in a vocabulary of 181 entity classes and 278 predicates: every row is a subject, a predicate and an object. The subject is an entity (a person, an app, an event); the object is either a plain value (a name, a handle, a date) or another entity, which is where the graph gets its edges. The predicate is the interesting part: it says what kind of relationship the system believes holds between the two.
an edge, and the rows it is stored as
One edge of the graph (synthetic data): a person and the organization the store holds as their employer. PS500 is the id predicate.json gives employer, and its range is not an organization but a relationship type — so one edge is two rows sharing a relationshipId, one naming the kind of relationship and one pointing at the other entity. The forty-six relationship type classes in the table below exist to carry exactly that, which is what lets an edge hold a confidence of its own and be retracted later.
Most of what the two daemons do is populate those predicates, and the tables kept beside them, from everything the OS will let them read. Into the graph go Contacts; Photos (faces); Calendar (titles and attendees); Family Sharing (including the age of every member); the Medical ID emergency contact in Health; events and contacts Siri Suggestions finds in Mail and Messages; people donated by apps through App Intents; and installed apps. Beside it, in tables of their own, sit who appears with whom in your photos; the communication log (who, when, which app, which direction; no content), which also pulls from your other devices; location and movement; and a long tail of device signals (screen locked, Wi-Fi, Bluetooth, charging, ambient light, and a stream named ContextualUnderstanding.PhysicalSocial whose record format lists people by name, phone and email, each with a start and an end time; it reads like a log of who was physically around you, though we never saw it filled in).
What is inferred
The complete list of the 181 entity classes:
the 181 entity classes in class.json
Class
Label
Group
Mechanism found?
SB56
entity
core
implemented
SB104
person
core
implemented
SB152
place
core
implemented
nm_topic
topic
core
implemented
SB65
event
core
implemented
nm_document
document
core
could not find mechanism
SB473
home
core
could not find mechanism
SB10
organization
core
implemented
SB2
group of humans
core
implemented
SB144
software
core
implemented
SB193
organism
core
implemented
SB198
agent
core
implemented
SB83
concept
core
implemented
SB400
occurrence
core
implemented
SB60
role
core
implemented
SB197
siribase type
core
implemented
nm_knowledgeSource
knowledge source
core
implemented
SB170
time
core
could not find mechanism
SB175
group
core
implemented
SB470
device primary user
person
could not find mechanism
SB199
musical artist
person
could not find mechanism
SB14
sports team
person
implemented
SB417
sports organization
person
could not find mechanism
SB115
sports league
person
could not find mechanism
SB185
athlete
person
could not find mechanism
SB21
person in the arts
person
could not find mechanism
SB184
musician
person
could not find mechanism
nm_screenTimeSegment
screen time segment
person
could not find mechanism
CS8
sex
attribute
might be implemented (unverified)
CS4
unit of measurement
attribute
implemented
CS25
type of profession
attribute
implemented
CS37
religion
attribute
could not find mechanism
SB74
language
attribute
could not find mechanism
SB18
political party
attribute
could not find mechanism
SB461
genre
attribute
implemented
SB537
mode of transportation
attribute
implemented
SB544
automobile model
attribute
implemented
SB418
make
attribute
implemented
SB399
content rating
attribute
could not find mechanism
SB48
populated place
place
implemented
SB101
state
place
might be implemented (unverified)
SB163
country
place
might be implemented (unverified)
SB370
time zone
place
might be implemented (unverified)
SB131
county
place
implemented
SB471
calendar
event
could not find mechanism
SB472
calendar event
event
implemented
SB513
activity event
event
implemented
SB71
sports game event
event
could not find mechanism
SB77
holiday event
event
implemented
SB514
transportation activity
activity
implemented
SB515
working activity
activity
implemented
SB516
meeting activity
activity
implemented
SB517
eating activity
activity
might be implemented (unverified)
SB518
watching TV activity
activity
might be implemented (unverified)
SB519
communicating activity
activity
implemented
SB520
shopping activity
activity
implemented
SB523
traveling activity
activity
might be implemented (unverified)
SB524
reading activity
activity
implemented
SB525
vacationing activity
activity
implemented
SB526
gaming activity
activity
implemented
SB527
socializing activity
activity
might be implemented (unverified)
SB528
waking up activity
activity
might be implemented (unverified)
SB529
going to bed activity
activity
might be implemented (unverified)
SB522
sleeping activity
activity
implemented
SB530
exercising activity
activity
implemented
SB521
mindfulness activity
activity
might be implemented (unverified)
SB531
on the phone activity
activity
implemented
SB532
facetime activity
activity
implemented
SB533
walking activity
activity
implemented
SB534
running activity
activity
implemented
SB535
bicycling activity
activity
implemented
SB536
hiking activity
activity
could not find mechanism
SB722
commute activity
activity
implemented
SB723
dining activity
activity
implemented
SB726
attending a performance activity
activity
implemented
SB732
stationary activity
activity
implemented
SB764
location visit activity
activity
implemented
SB784
driving activity
activity
could not find mechanism
SB786
behavior activity
activity
could not find mechanism
SB394
sports activity
activity
implemented
SB116
type of activity
activity type
might be implemented (unverified)
SB545
transportation activity type
activity type
could not find mechanism
SB546
working activity type
activity type
could not find mechanism
SB547
meeting activity type
activity type
could not find mechanism
SB548
eating activity type
activity type
could not find mechanism
SB549
watching media activity type
activity type
could not find mechanism
SB550
communicating activity type
activity type
could not find mechanism
SB551
shopping activity type
activity type
could not find mechanism
SB552
mindful activity type
activity type
could not find mechanism
SB553
sleep activity type
activity type
could not find mechanism
SB554
travel activity type
activity type
could not find mechanism
SB555
reading activity type
activity type
could not find mechanism
SB556
vacation activity type
activity type
could not find mechanism
SB557
social activity type
activity type
could not find mechanism
SB558
waking activity type
activity type
could not find mechanism
SB559
going to bed activity type
activity type
could not find mechanism
SB560
phone call activity type
activity type
could not find mechanism
SB561
facetime activity type
activity type
could not find mechanism
SB562
walking activity type
activity type
could not find mechanism
SB563
running activity type
activity type
could not find mechanism
SB564
biking activity type
activity type
could not find mechanism
SB565
hiking activity type
activity type
could not find mechanism
SB483
type of physical exercise
activity type
could not find mechanism
SB721
commute activity type
activity type
could not find mechanism
SB785
driving activity type
activity type
could not find mechanism
SB787
behavior activity type
activity type
could not find mechanism
SB425
vehicle
transport
implemented
SB543
on foot
transport
could not find mechanism
SB335
automobile
transport
implemented
SB538
plane
transport
could not find mechanism
SB539
bicycle
transport
could not find mechanism
SB358
ship
transport
could not find mechanism
SB541
bus
transport
could not find mechanism
SB542
locomotive train
transport
could not find mechanism
SB51
work
media
implemented
SB405
media work
media
implemented
SB44
song
media
implemented
SB138
album
media
could not find mechanism
SB85
musical work
media
implemented
SB183
movie
media
could not find mechanism
SB295
television program
media
could not find mechanism
SB49
animal
animal
could not find mechanism
nm_pet
pet
animal
could not find mechanism
nm_dog
dog
animal
could not find mechanism
nm_cat
cat
animal
could not find mechanism
nm_myPet
my pet
animal
could not find mechanism
nm_myDog
my dog
animal
could not find mechanism
nm_myCat
my cat
animal
could not find mechanism
nm_walletOrder
wallet order
commerce
implemented
nm_walletTransaction
wallet transaction
commerce
could not find mechanism
nm_payment
payment
commerce
implemented
nm_transaction
transaction
commerce
implemented
nm_email
commerce
implemented
nm_shippingInfo
shippingInfo
commerce
could not find mechanism
nm_shipmentFullFillmentInfo
shipment full fillment info
commerce
implemented
CS29
location relationship type
edge
implemented
CS97
address relationship type
edge
implemented
nm_topicRelationshipType
topic relationship type
edge
might be implemented (unverified)
CS1
relationship type
edge
implemented
nm_senderRelationshipType
sender relationship type
edge
might be implemented (unverified)
nm_calendarRelationshipType
calendar relationship type
edge
implemented
nm_organizerRelationshipType
organizer relationship type
edge
could not find mechanism
CS105
text relationship type
edge
could not find mechanism
CS101
language relationship type
edge
could not find mechanism
CS22
identifier relationship type
edge
implemented
CS2
image relationship type
edge
might be implemented (unverified)
CS13
membership relationship type
edge
implemented
CS165
employment relationship type
edge
implemented
CS154
time zone relationship type
edge
might be implemented (unverified)
CS3
quantity relationship type
edge
implemented
CS36
religion relationship type
edge
could not find mechanism
CS167
interest relationship type
edge
could not find mechanism
CS38
participation relationship type
edge
implemented
nm_calendarSourceRelationshipType
calendar source relationship type
edge
implemented
CS168
agent affiliation relationship type
edge
implemented
nm_dateRelationshipType
date relationship type
edge
implemented
nm_administratorRelationshipType
administrator relationship type
edge
could not find mechanism
nm_visualIdentifierRelationshipType
visual identifier relationship type
edge
implemented
nm_songRelationshipType
song relationship type
edge
implemented
nm_softwareRelationshipType
software relationship type
edge
implemented
CS173
event relationship type
edge
could not find mechanism
nm_localizedNameRelationshipType
localized name relationship type
edge
implemented
nm_confidenceRelationshipType
confidence relationship type
edge
implemented
CS31
performance relationship type
edge
could not find mechanism
CS15
part whole relationship type
edge
could not find mechanism
CS113
rating relationship type
edge
could not find mechanism
CS166
group relationship type
edge
implemented
nm_livingBeingRelationship
living being relationship
edge
implemented
nm_entityAliasRelationshipType
entity alias relationship type
edge
implemented
nm_extractionTagRelationshipType
extraction tag relationship type
edge
implemented
nm_orderEmailRelationshipType
order email relationship type
edge
implemented
nm_orderTransactionRelationshipType
order transaction relationship type
edge
implemented
nm_orderShippingInfoRelationshipType
order shipping info relationship type
edge
implemented
nm_entityOrganizationRelationshipType
entity organization relationship type
edge
implemented
nm_shipmentFullFillmentEmailRelationshipType
shipment full fillment email relationship type
edge
implemented
nm_orderShipmentFullFillmentRelationshipType
order shipment full fillment relationship type
edge
implemented
nm_orderPaymentRelationshipType
order payment relationship type
edge
implemented
nm_fromToEmailRelationshipType
from to email relationship type
edge
could not find mechanism
nm_orderPersonRelationshipType
order person relationship type
edge
implemented
nm_shipmentPersonRelationshipType
shipment person relationship type
edge
implemented
nm_emailInfoRelationshipType
email info relationship type
edge
could not find mechanism
Every class the shipped ontology declares, grouped rather than left in file order: the core types, then people, attributes, places and events, then the activity block, and last the forty-six relationship types, which are not things at all — each one reifies an edge so that the edge can carry facets of its own, a confidence among them. The ids and labels are Apple's; the grouping is ours. intValue runs 1 to 185 with four numbers missing — 18, 54, 169, 181 — and nothing on the disk says what they were.
Fifty-seven of the 181 describe an activity: sleeping, exercising, mindfulness, going to bed, commute, dining, most of them with a matching type class beside them, so nearly a third of the vocabulary went on what you are doing rather than on who or where you are. That said, it is not clear that every class is actually used; some may be placeholders that are declared and never filled. In the table, we flag as implemented the classes for which we could locate a mechanism that might be responsible for the class, but those are to be taken as inference work rather than ground truth.
And this is the complete list of predicates (same disclaimer applies):
the 278 predicates in predicate.json
Predicate
What it records
Data source
How it is obtained
Mechanism found?
PS1
what kind of thing this is
every source
Rule
implemented
PS10
genre
Apple world knowledge
Apple-shipped
implemented
PS102
this record is the same person/thing as that one
every source
Embedding model
implemented
PS106
duration
Calendar, Suggestions
Copied
could not find mechanism
PS107
has location
Calendar, location history
Copied
might be implemented (unverified)
PS123
part in part-whole relationship
Apple world knowledge
Apple-shipped
could not find mechanism
PS132
has religion
Contacts
Copied
could not find mechanism
PS133
religion
Apple world knowledge
Apple-shipped
could not find mechanism
PS135
area
Apple world knowledge
Apple-shipped
could not find mechanism
PS137
performed by
Apple world knowledge
Apple-shipped
could not find mechanism
PS143
attendee identifier
Calendar, Suggestions
Copied
might be implemented (unverified)
PS144
took part in an event
Calendar, Photos
Copied
could not find mechanism
PS147
population
Apple world knowledge
Apple-shipped
could not find mechanism
PS16
performer
Apple world knowledge
Apple-shipped
could not find mechanism
PS194
published by
Apple world knowledge
Apple-shipped
implemented
PS2
unit
Apple world knowledge
Apple-shipped
implemented
PS225
has location relationship
Calendar, location history
Copied
implemented
PS232
has attendee
Calendar, Suggestions
Copied
implemented
PS248
starting location
Calendar, location history
Copied
implemented
PS249
end location
Calendar, location history
Copied
implemented
PS29
start time
Calendar, Suggestions
Copied
implemented
PS296
URL
Calendar, apps, world knowledge
Copied
implemented
PS3
quantity
Apple world knowledge
Apple-shipped
implemented
PS30
end time
Calendar, Suggestions
Copied
implemented
PS312
first name
Contacts
Copied
implemented
PS313
username / handle
Contacts, Family Sharing
Copied
implemented
PS314
sport
Apple world knowledge
Apple-shipped
could not find mechanism
PS315
postal code
Contacts, order emails
Copied
implemented
PS316
full street address
Contacts, order emails
Copied
might be implemented (unverified)
PS317
has address
Contacts, order emails
Copied
implemented
PS324
has language
Contacts
Copied
could not find mechanism
PS325
languages spoken
Contacts
Copied
could not find mechanism
PS33
name
Contacts
Copied
implemented
PS335
text value
Calendar, apps, world knowledge
Copied
could not find mechanism
PS336
official name
Apple world knowledge
Apple-shipped
could not find mechanism
PS34
gender identity
Contacts
Copied
might be implemented (unverified)
PS343
demonym
Apple world knowledge
Apple-shipped
could not find mechanism
PS35
citizen of
Apple world knowledge
Apple-shipped
could not find mechanism
PS358
has rating
Apple world knowledge
Apple-shipped
could not find mechanism
PS359
rating
Apple world knowledge
Apple-shipped
could not find mechanism
PS36
birth name
Contacts
Copied
implemented
PS37
date of birth
Contacts
Copied
implemented
PS374
has member
Contacts, Family Sharing
Copied
implemented
PS392
last updated
Calendar, Suggestions
Copied
implemented
PS396
external identifier, its value, its type
every source
Joined
implemented
PS4
width
Apple world knowledge
Apple-shipped
might be implemented (unverified)
PS406
email address
Contacts, order emails
Copied
implemented
PS407
phone number
Contacts, order emails
Copied
implemented
PS424
member of political party
Apple world knowledge
Apple-shipped
could not find mechanism
PS441
family name
Contacts
Copied
implemented
PS447
has timezone
Calendar, location history
Copied
could not find mechanism
PS448
time zone
Calendar, location history
Copied
might be implemented (unverified)
PS45
member of
Contacts, Family Sharing
Copied
implemented
PS450
address country
Contacts, order emails
Copied
might be implemented (unverified)
PS453
city
Contacts, order emails
Copied
might be implemented (unverified)
PS454
country code ISO 3166-1 alpha-2
Contacts, order emails
Copied
might be implemented (unverified)
PS455
administrative district
Contacts, order emails
Copied
might be implemented (unverified)
PS46
group member identifier
Family Sharing
Joined
could not find mechanism
PS484
official language
Apple world knowledge
Apple-shipped
could not find mechanism
PS486
display image
Photos, Contacts
Copied
could not find mechanism
PS5
height
Apple world knowledge
Apple-shipped
might be implemented (unverified)
PS500
employer
Contacts
Copied
implemented
PS51
description
Calendar, apps, world knowledge
Copied
implemented
PS515
location within location
Apple world knowledge
Apple-shipped
could not find mechanism
PS520
has contact information
Contacts, order emails
Copied
implemented
PS524
native home
Apple world knowledge
Apple-shipped
could not find mechanism
PS53
part of
Apple world knowledge
Apple-shipped
could not find mechanism
PS545
name prefix
Contacts
Copied
implemented
PS546
middle name
Contacts
Copied
implemented
PS547
name suffix
Contacts
Copied
implemented
PS548
department
Contacts
Copied
implemented
PS557
topics a person is interested in
Personalization Portrait
Scored model
could not find mechanism
PS558
topics a person is interested in
Personalization Portrait
Scored model
could not find mechanism
PS565
abbreviation
Apple world knowledge
Apple-shipped
could not find mechanism
PS566
group has part
Apple world knowledge
Apple-shipped
could not find mechanism
PS567
seconds from GMT
Calendar, location history
Copied
could not find mechanism
PS568
sub administrative area
Contacts, order emails
Copied
implemented
PS569
sub locality
Contacts, order emails
Copied
implemented
PS570
sub thoroughfare
Contacts, order emails
Copied
implemented
PS571
latitude
Calendar, location history
Copied
implemented
PS572
longitude
Calendar, location history
Copied
implemented
PS573
has latitude/longitude
Calendar, location history
Copied
implemented
PS574
previous family name
Contacts
Copied
implemented
PS578
has organizer
Calendar, Suggestions
Copied
implemented
PS579
on calendar
Calendar, Suggestions
Copied
could not find mechanism
PS580
calendar event sender
Calendar, Suggestions
Copied
implemented
PS581
second address line
Contacts, order emails
Copied
implemented
PS582
who administers the home
HomeKit
Copied
could not find mechanism
PS586
status
—
Not established
implemented
PS598
a person is related to / associated with another
Contacts, Health, Family Sharing, Screen Time, Photos
XGBoost + rules + copied
implemented
PS599
a person is related to / associated with another
Contacts, Health, Family Sharing, Screen Time, Photos
XGBoost + rules + copied
could not find mechanism
PS618
has subevent
Calendar, Suggestions
Copied
could not find mechanism
PS66
source URL
Calendar, apps, world knowledge
Copied
implemented
PS67
date of death
Apple world knowledge
Apple-shipped
could not find mechanism
PS69
external identifier, its value, its type
every source
Joined
implemented
PS7
image file
Photos, Contacts
Copied
might be implemented (unverified)
PS709
has driver
CarPlay, Wallet
Copied
could not find mechanism
PS710
transportation type
Apple world knowledge
Apple-shipped
could not find mechanism
PS711
elevation gain
Apple world knowledge
Apple-shipped
could not find mechanism
PS712
terrain
Apple world knowledge
Apple-shipped
implemented
PS713
fuel consumed
Apple world knowledge
Apple-shipped
could not find mechanism
PS714
highway mileage
Apple world knowledge
Apple-shipped
could not find mechanism
PS715
city mileage
Apple world knowledge
Apple-shipped
could not find mechanism
PS716
make of automobile
CarPlay, Wallet
Copied
could not find mechanism
PS717
model of automobile
CarPlay, Wallet
Copied
could not find mechanism
PS718
two events happened together
Photos, Calendar
Rule
could not find mechanism
PS719
kind of activity
Health, motion sensors
Rule
could not find mechanism
PS72
external identifier, its value, its type
every source
Joined
implemented
PS73
occupation
Contacts, Apple world knowledge
Copied
implemented
PS730
sub premises
Contacts, order emails
Copied
might be implemented (unverified)
PS731
administrative area code
Contacts, order emails
Copied
might be implemented (unverified)
PS733
label the user gave a place or thing
Calendar, Suggestions
Copied
might be implemented (unverified)
PS734
participant obligation
Calendar, Suggestions
Copied
could not find mechanism
PS740
contact label
Contacts
Copied
implemented
PS741
has event
Calendar, Suggestions
Copied
could not find mechanism
PS747
behavior category
Behavioral history
Rule
could not find mechanism
PS75
has mother
—
Not established
could not find mechanism
PS753
group includes individual
Family Sharing
Copied
implemented
PS76
has father
—
Not established
could not find mechanism
PS77
spouse
—
Not established
could not find mechanism
PS78
has image
Photos, Contacts
Copied
could not find mechanism
PS79
has child
—
Not established
could not find mechanism
PS83
length
Apple world knowledge
Apple-shipped
could not find mechanism
PS85
role held
Apple world knowledge
Apple-shipped
could not find mechanism
PS89
another name the same person is known by
Contacts, Photos
Embedding model
implemented
PS99
net worth
Apple world knowledge
Apple-shipped
could not find mechanism
nm_QID
Wikidata id
Apple world knowledge
Apple-shipped
implemented
nm_UMC_ID
UMC ID
Apple Music
Copied
implemented
nm_accountInstitution
account institution
Wallet, order emails
Copied
implemented
nm_accountLastFourDigits
account last four digits
Wallet, order emails
Copied
implemented
nm_accountType
account type
Wallet, order emails
Copied
implemented
nm_accuracy
accuracy
Location history
Copied
could not find mechanism
nm_activityCategory
event type pulled out of text
Mail, Messages, app content
Rule
could not find mechanism
nm_addressString
address string
Calendar, Suggestions
Copied
implemented
nm_age
age
Family Sharing
Copied
implemented
nm_albumIdentifier
album identifier
Apple Music
Copied
implemented
nm_albumName
song album name
Apple Music, app usage
Copied
implemented
nm_amount
amount
Wallet, order emails
Copied
implemented
nm_anniversary
anniversary
Contacts
Copied
implemented
nm_applicationIdentifier
application identifier
Installed apps
Copied
implemented
nm_artistIdentifier
artist identifier
Apple Music
Copied
implemented
nm_artistName
song artist name
Apple Music, app usage
Copied
implemented
nm_associationReason
which mechanism named this face
Photos faces
Rule
implemented
nm_availability
availability
Calendar, Suggestions
Copied
could not find mechanism
nm_birthdayContactID
birthday contact id
Contacts
Joined
could not find mechanism
nm_bundleID
bundle id
Installed apps
Copied
implemented
nm_calendarType
calendar type
Calendar, Suggestions
Copied
could not find mechanism
nm_carrierName
carrier name
Wallet passes, Calendar
Copied
implemented
nm_category
topic scored from what you read and write
Personalization Portrait
Scored model
could not find mechanism
nm_cnFavoriteActionType
contact favorite action type
Contacts
Copied
implemented
nm_confidence
how sure the system is
whichever engine wrote the fact
Rule
implemented
nm_confirmationConfidence
how sure a name match is
Contacts, Photos
Embedding model
implemented
nm_creationDate
creation date
Calendar, Suggestions
Copied
implemented
nm_dates
dates
Calendar, Suggestions
Copied
implemented
nm_documentID
document ID
—
Not established
could not find mechanism
nm_dynamicCategory
topic scored from what you read and write
Personalization Portrait
Scored model
could not find mechanism
nm_emailFrom
email from
Order emails
Copied
implemented
nm_emailIdentifier
email identifier
Order emails
Joined
implemented
nm_emailReplyTo
email reply to
Order emails
Copied
implemented
nm_emailTo
email to
Order emails
Copied
could not find mechanism
nm_emailType
email type
Order emails
Copied
implemented
nm_entityAliasRelationship
link joining two names for one person
Contacts, Photos
Embedding model
implemented
nm_exactVersion
exact version
Installed apps
Copied
could not find mechanism
nm_extensionHostId
extension host id
Installed apps
Copied
could not find mechanism
nm_familyMemberType
family member type
Family Sharing
Copied
implemented
nm_file
file
Installed apps
Copied
could not find mechanism
nm_flightNumber
flight number
Wallet passes, Calendar
Copied
implemented
nm_handleIdentifier
handle identifier
Messages, Mail
Joined
implemented
nm_handles
the phone numbers and emails a person uses
Contacts
Copied
might be implemented (unverified)
nm_hasAdministrator
has administrator
HomeKit
Copied
could not find mechanism
nm_hasCalendar
has calendar
Calendar, Suggestions
Copied
could not find mechanism
nm_hasConfidence
per-source trust weight
Downloadable Apple asset
Apple-shipped
implemented
nm_hasDate
has date
Calendar, Suggestions
Copied
implemented
nm_hasDomain
which classes a predicate applies to
The ontology itself
Apple-shipped
could not find mechanism
nm_hasEmail
has email
Order emails
Copied
implemented
nm_hasExtractionTags
has extraction tags
Ingest lane
Rule
could not find mechanism
nm_hasFavoriteSportsTeam
has favorite sports team
App usage, Apple Music
Copied
could not find mechanism
nm_hasField
has field
—
Not established
implemented
nm_hasLocalizedName
has localized name
Apple world knowledge
Apple-shipped
implemented
nm_hasMerchant
has merchant
Wallet, order emails
Copied
implemented
nm_hasPayment
has payment
Wallet, order emails
Copied
implemented
nm_hasPerson
has person
Order emails
Copied
implemented
nm_hasProfileImage
has profile image
Contacts
Copied
implemented
nm_hasRange
which classes a predicate applies to
The ontology itself
Apple-shipped
implemented
nm_hasSender
has sender
Order emails
Copied
might be implemented (unverified)
nm_hasShipmentFullFillment
has shipment full fillment
Wallet, order emails
Copied
implemented
nm_hasShippingInfo
has shipping info
Wallet, order emails
Copied
could not find mechanism
nm_hasSoftware
has software
Installed apps
Copied
implemented
nm_hasSong
has song
Apple Music, app usage
Copied
implemented
nm_hasSource
has source
Calendar, Suggestions
Copied
could not find mechanism
nm_hasTopic
topic scored from what you read and write
Personalization Portrait
Scored model
might be implemented (unverified)
nm_hasTransaction
has transaction
Wallet, order emails
Copied
implemented
nm_hasVisualIdentifier
the face cluster for a person
Photos faces
Face clustering
implemented
nm_homeTeam
home team
Apple Music, app usage
Copied
could not find mechanism
nm_iataCode
iata code
Wallet passes, Calendar
Copied
implemented
nm_image_bytes
image bytes
Photos, Contacts
Copied
implemented
nm_imputedAnniversary
imputed anniversary
Contacts, behavior
Unknown model
implemented
nm_imputedBirthday
birthday filled in rather than told — and the one the system actually reads
Photos age estimate (candidate)
Unknown model
implemented
nm_imputedCreationDate
timestamp filled in or normalized
Calendar, Suggestions
Rule
implemented
nm_imputedEndTime
timestamp filled in or normalized
Calendar, Suggestions
Rule
implemented
nm_imputedInferredBirthday
birthday filled in rather than told — and the one the system actually reads
Photos age estimate (candidate)
Unknown model
implemented
nm_imputedLastUpdated
timestamp filled in or normalized
Calendar, Suggestions
Rule
implemented
nm_imputedNonGregorianBirthday
imputed non-gregorian birthday
Contacts, behavior
Unknown model
implemented
nm_imputedOccurrenceDate
timestamp filled in or normalized
Calendar, Suggestions
Rule
implemented
nm_imputedStartTime
timestamp filled in or normalized
Calendar, Suggestions
Rule
implemented
nm_inferenceFeedbackID
handle marking a machine-made claim
The kinship classifier
XGBoost
implemented
nm_inferredBirthday
birthday filled in rather than told — and the one the system actually reads
Photos age estimate (candidate)
Unknown model
implemented
nm_inferredDeviceUser
owner worked out rather than declared
Behavioral history
Unknown model
implemented
nm_inferredSemanticLabel
what a place is for, e.g. home, gym
Location history
Unknown model
implemented
nm_intentPersonId
person as an app referred to them
App donations
Joined
implemented
nm_isAllDay
is all day
Calendar, Suggestions
Copied
implemented
nm_isCurrentUser
this person is the phone's owner
Contacts me-card
Rule
implemented
nm_isDetached
is detached
Calendar, Suggestions
Copied
could not find mechanism
nm_isFavorite
is favorite
Contacts
Copied
implemented
nm_isStarting
is starting
Calendar, Suggestions
Copied
could not find mechanism
nm_isSubscribed
subscribed status
Calendar, Suggestions
Copied
could not find mechanism
nm_launchReason
launch reason
App usage, Screen Time
Copied
could not find mechanism
nm_launchType
launch type
App usage, Screen Time
Copied
could not find mechanism
nm_league_name
league name
Apple Music, app usage
Copied
implemented
nm_listening_confidence
per-language competence on a relationship
—
Not established
might be implemented (unverified)
nm_localeIdentifier
locale identifier
System settings
Copied
implemented
nm_mappedTopic
topic scored from what you read and write
Personalization Portrait
Scored model
could not find mechanism
nm_merchantName
merchant name
Wallet, order emails
Copied
implemented
nm_naturalLanguageEventTypeIdentifier
event type pulled out of text
Mail, Messages, app content
Rule
could not find mechanism
nm_nonGregorianBirthday
non-gregorian birthday
Contacts
Copied
implemented
nm_occurrenceDate
occurrence date
Calendar, Suggestions
Copied
implemented
nm_orderContentType
order content type
Wallet, order emails
Copied
implemented
nm_orderNumber
order number
Wallet, order emails
Copied
implemented
nm_organizationIdentifier
organization identifier
Contacts, Wallet
Joined
implemented
nm_parentBundleId
parent bundle id
Installed apps
Copied
could not find mechanism
nm_paymentIdentifier
payment identifier
Wallet
Joined
implemented
nm_paymentMethodIsApplePay
payment method is apple pay
Wallet, order emails
Copied
implemented
nm_personIdentifier
who a relationship points at
the other end of a relationship
Joined
implemented
nm_personType
what kind of thing Photos detected
Photos
Rule
implemented
nm_phoneticFamilyName
phonetic family name
Contacts
Copied
implemented
nm_phoneticFirstName
phonetic first name
Contacts
Copied
implemented
nm_phoneticMiddleName
phonetic middle name
Contacts
Copied
implemented
nm_phoneticName
phonetic name
Contacts
Copied
implemented
nm_photosDetectionType
what kind of thing Photos detected
Photos
Rule
implemented
nm_poiType
kind of place
Maps, location history
Copied
could not find mechanism
nm_reading_confidence
per-language competence on a relationship
—
Not established
might be implemented (unverified)
nm_relationship
a person is related to / associated with another
Contacts, Health, Family Sharing, Screen Time, Photos
XGBoost + rules + copied
implemented
nm_relationshipLabel
the relationship word, e.g. sister, coworker
Contacts, Health, Family Sharing
XGBoost + copied
implemented
nm_role
role
—
Not established
could not find mechanism
nm_screenTimeSegmentPerson
screen time segment person
Screen Time
Copied
could not find mechanism
nm_senderDomain
sender domain
Order emails
Copied
implemented
nm_senderIdentifier
sender identifier
Messages, Mail
Joined
might be implemented (unverified)
nm_sgEventType
event type worked out from an app or email
Suggestions
Rule
implemented
nm_shipmentFullFillmentIdentifier
shipment full fillment identifier
Wallet
Joined
implemented
nm_shippingMethod
shipping method
Wallet, order emails
Copied
implemented
nm_shippingStatus
shipping status
Wallet, order emails
Copied
implemented
nm_shortVersion
short version
Installed apps
Copied
could not find mechanism
nm_signalLabel
signal label
Photos
Rule
implemented
nm_softwareIdentifier
software identifier
Installed apps
Joined
might be implemented (unverified)
nm_songIdentifier
song identifier
Apple Music
Copied
could not find mechanism
nm_sourceType
source type
Ingest lane
Rule
implemented
nm_subject
subject
Order emails
Copied
implemented
nm_suggestedName
name guessed for an unnamed face
Photos faces + Messages text
Local LLM
implemented
nm_suggestedRelationshipConfidence
guessed relationship of a face to the phone's owner
Photos faces + Messages text
Local LLM
implemented
nm_suggestedRelationshipToDeviceOwner
guessed relationship of a face to the phone's owner
Photos faces + Messages text
Local LLM
implemented
nm_teamIdentifier
team identifier
Installed apps
Joined
implemented
nm_toDisplayName
display name on an email recipient
Copied
implemented
nm_toEmailAddress
to email address
Copied
implemented
nm_topicIdentifier
topic identifier
Personalization Portrait
Joined
might be implemented (unverified)
nm_totalAmount
total amount
Wallet, order emails
Copied
implemented
nm_totalNotifications
total notifications
App usage, Screen Time
Copied
could not find mechanism
nm_totalPickups
total pickups
App usage, Screen Time
Copied
could not find mechanism
nm_trackingNumber
tracking number
Wallet, order emails
Copied
implemented
nm_transactionIdentifier
transaction identifier
Wallet
Joined
implemented
nm_transactionSource
transaction source
Wallet, order emails
Copied
implemented
nm_transactionStatus
transaction status
Wallet, order emails
Copied
implemented
nm_transactionType
transaction type
Wallet, order emails
Copied
implemented
nm_uncertainty
uncertainty
Location history
Copied
could not find mechanism
nm_unknownHandle
unknown handle
Messages, Mail
Joined
implemented
nm_unlocalizedRelationshipLabel
the relationship word, e.g. sister, coworker
Contacts, Health, Family Sharing
XGBoost + copied
implemented
nm_virtualConferenceURL
virtual conference URL
Calendar, Suggestions
Copied
implemented
nm_visualIdentifier
the face cluster for a person
Photos faces
Face clustering
implemented
nm_writing_confidence
per-language competence on a relationship
—
Not established
might be implemented (unverified)
Every predicate the shipped ontology declares. The first column is the id as graph.db stores it; the last says how a value gets there: copied from a source, joined from another record, derived by a rule, shipped by Apple as world knowledge, or produced by a model (in bold). Seven we could not establish, and the table says so rather than guessing. The ids and labels are Apple's; the plain-language readings are ours.
How it is inferred
The methods run the whole range. Most of the vocabulary is filled by plain copying (a name from Contacts, an attendee from Calendar) and rules over the copies; the interesting rows come from models: a pair of XGBoost classifiers decides who is family from 144 mostly behavioral features, two embedding models are one of ten signals the system uses to decide when two records are the same person, twelve analyzers in Photos score partner, parent, child, coworker and friend from who appears in your pictures with you, on which trips and holidays, and Apple's on-device LLM reads your messages to put a name and a relationship on a face (if Apple Intelligence is active).
Whatever the method, every row lands in the graph with a confidence attached, a number between 0 and 1.
The family scores live apart. Two CoreML models ship next to the ontology, EntityTagging_Family.mlmodelc and EntityTagging_FamilyAndFriends.mlmodelc: binary XGBoost classifiers that take 144 features and return a class and a probability over it. Their scores do not go into the triple store but into a table of their own (entityTagging.db, one row per person, seventeen relationship columns from family to mother to coworker, recomputed every day), which is what the iKnow scorecard further down reads. Only two of those seventeen columns have a shipped model behind them; what fills the other fifteen, and how any of these probabilities would become a confidence in the graph, we could not trace.
The feature names are in each model's metadata.json, in plain text. They are the clearest statement of what this system is that we have found anywhere. Sixteen of the 144:
Sixteen of the 144 inputs EntityTagging_Family.mlmodelc reads, numbered by their place in its metadata.json: the six this post names, and ten that show how the rest are cut (overall, by day, by time of day, by place). EntityTagging_FamilyAndFriends reads the same 144 in the same order, and every one is a number. The groups are the schema's own three blocks: twenty on who the person is, forty on when and how you communicate with them, eighty-four on the photographs you both appear in. Two of the twenty, photosPersonAge and photosAgeDifferenceCategory, are read from Photos too. The names are Apple's; the group labels are ours.
The identity group is the tidy story: same surname, same home address, an iCloud Family flag — and it is the small half. The bulk of the model is the other two: whom you message and who appears in your photographs, conditioned on the day of the week and the time of day, and then on both at once. photosPersonPosteriorProbabilityGivenSaturdayNight is a shipped feature name in a shipped classifier whose job is deciding whether someone is family.
Deleting something does not delete it everywhere.
The live graph forgets. Delete a contact and nothing re-derives them; the next run replaces the table wholesale.
But the store keeps a second table, expired_stable_graph, beside the live one. The retracted row moves there and keeps its confidence and its sources. How long it stays there is not clear from what we could see. It seems to go after about 48 hours (the store keeps a prune watermark, and a recent snapshot of the table reached back no further than that); but on our own Mac, we could still read very old records in it.
And a conclusion can outlive the evidence for it. On our Mac, the scorecard still scored two people the graph had dropped the day before, both of them known to it only from a calendar invitation.
Look into your own data
Is any of this sensitive? We'll leave that to you, since "sensitive" means different things to different people (though we think you have plenty to choose from here). However, the thing is, none of it lives in isolation. It is built to be joined and correlated; the nodes are there to be linked:
a personal knowledge graph (synthetic data)
me · the owner of this Mac · 1a group the store holds as an entity · 1family, as scored · 7scored close to me · 9employer · 4group chat · 6known, not scored close · 61membership of a group chata calendar event both attendedan employment statementnamed by the stored groupa closeness score above the cut · nobody typed it
How to read this
Every node is a row in the store, and the record under it is what the store holds. Size is how many rows name it; shape is what kind of thing it is; the one node in accent is you, and a ring marks the four the store scores as family. A dashed line is a conclusion nobody typed.
Click a node — or reach the graph with Tab, walk it with the arrow keys and press Enter — and its record replaces these notes.
Click a node and the record under the graph is what the store holds on it (the rows, not a summary of them). The sixty-one faint ones are the part nobody expects: people the store knows and has not placed, in it because they wrote to you once, or because they were on somebody else's invitation.
The most important point, though, is that this is not only about you (the person using the Mac). The people around you are in it too: your family, your friends, "anyone" you have talked to through your device. None of them was asked, and the store links them to you and, through the events and groups they share, to each other. A relationship is intelligence one level above the rows it was inferred from. For instance, by correlating across our own store, we (well, an LLM working for us) could easily work out the following:
That there is a child in the household, and roughly how old. The family-member types and a family circle give the household. The scorecard's child and son/daughter columns give the child's place in it. Beside it sits an age, copied from Family Sharing; and dates of birth in the store can carry the system's own word imputed, though what the system means by it is not clear to us.
Which daycare the child goes to, and both of the numbers that reach it. Copied off contact cards. Note what is not there: the calendar has no school-run event, so the store holds the institution and its lines but not the hour the child is dropped off. That is still enough to call a parent from a number their phone already trusts.
Where the family will be, on which days, between which hours. One calendar event, a place, two stored time slots and a title that names the family (so the guest list is in the string even where the attendee rows are empty).
Which of the family will be there. The scorecard names the ones scored as family, a household entity holds them together, and a Family Sharing circle says which of them are in the same family group.
And, in the same file, what is installed on the machine, down to the password manager.
It is easy to see how this could be weaponized if this single point of failure for your privacy ends up in the wrong hands. We play that (hypothetical) scenario out in a second write-up. The good thing is that this knowledge is meant to stay local on your machine, protected by the Full Disk Access permission; although it can be consumed by other services (but it is not entirely clear to us how and what can be accessed).
The reality is that the amount of information in this store is really something else: every time we dig into what the daemons infer, and how, a few new lines of investigation open up. Whatever you would rather no one knew about you or your family may well be in there somewhere, and we think you have the right to know whether it is. What you get instead: no documentation, no way to inspect the store and no general way to correct an inference that is wrong.
So we give you two tools. The first is iKnow, a local dashboard for browsing your own graph without anything leaving your Mac (the next section walks through it). The second is for going deeper. The iKnow repository ships a manual for your own agent that tells it where the databases are and how to read them, so it can go looking for whatever you are after. (Please point a self-hosted model at it, not a cloud one.)
iKnow
We designed iKnow as a web dashboard implemented in the Python standard library; no npm and no CDN.
It is meant to be run directly on your Mac, where the databases are stored. When executed, it binds 127.0.0.1, locates the databases in your file system (granted the necessary permissions), opens them read-only, collects the info on the fly and represents it on a dashboard. Nothing is written and nothing is logged; values go from the stores, over loopback, into a browser tab; it can be run air-gapped if you want.
We have tested it on macOS 26.6.2 and 26.7.1, and, only partially, on 27.0.1.
iKnow has a dozen pages.
Start here
What your Mac knows. The landing page: the graph of people around you. Below it come the faces in your photos, your family, your handles, your correspondents and where you will be next.
Its graph is the same household as the one at the top of this post, this time drawn by the tool's own renderer:
the landing page
The first screen. The white node marked (Me) is the owner of the Mac; the rest are the people, employers and group chats the store links to it, colored by what it says each one is. Orange is family, as scored. Blue is a closeness score above the cut. The faded names around the edges are people the store knows and has not placed.
Nothing here is summarized that is not in the record: every node is a row, and clicking one shows you the row (which is what the graph at the top of this post does too).
The graph is the overview, and an overview is easy to shrug at. The pages behind it are not.
Social graph. Whom the store links to whom through shared meetings, group chats and a closeness score. Each circle opens to its members and to the evidence behind it.
What it concluded
Kinship scorecard. The family bindings inferred by the daemon.
Its card on the landing page is cut down to the people it names:
Family, as the scorecard has it
Six people, six relationships, and a subtitle saying nobody typed any of it. That holds for the scores; some of what they are computed from, you did type (a relation on a contact card, an emergency contact, Family Sharing). A mother, a father and a daughter at 0.960, which is as high as these columns go; a brother, a son and a sister at 0.500.
Below that is the part we find hardest to argue with. Nine separate checks (the partner column, the composite score, the household group entity, the Family Sharing token, the iCloud Family flag, top twelve-month contact, a Photos face cluster, a Contacts record, a reachable handle), and the card counts how many of them land on the same person. Eight of nine on the mother; seven on the father, and seven on Priya Sandoval, the partner, who is not in the list above: iKnow lists the relationships that roll up into another column (mother and father into parent), and partner rolls up into nothing. The join found the partner anyway.
The card is a summary of one table, and iKnow draws the whole table too:
Every person, every column
Sorted by familyAndFriends, the broadest of the scores. The seven at the top each light up their own relationship and the column it rolls up into (mother and father into parent, sister and brother into sibling, son and daughter into child), and the partner is also a colleague. Further down are the friends and the rest of the office. Under them, almost everyone else is the same two cells: one repeated value on family and another on familyAndFriends, which is what the classifier writes when it has nothing to say. Two columns (alumni and unknown) are never used at all.
The scores in these two shots are invented, and more generous than the real thing. On the Mac we took apart, only mother reached the top of the ladder; the few others it scored at all got one small vote on every family label at once, father and brother alike. What the fixture keeps is the table's grain: it only ever writes a handful of values (0.125, 0.25, 0.5 and 0.96), and most people sit on one default.
The face index. A face cluster for everyone in your photos.
Its co-occurrence graph does not use a single contact, message or calendar entry. It is built from your photographs alone:
The face co-occurrence graph
Two faces are linked when they appear in the same photograph, and that is the only input: 8,904 clustered faces over the 1,643 photographs that hold more than one. The labels are fragments of each cluster's local identifier, because there are no names to show. The owner is the hub. The knot on the left is the family, the group at the bottom right is work, and the two never share a frame.
Each node wears its cluster's key face, read out of the photo library and blurred by iKnow's own switch for sharing the screen. Under the blur there is nobody: the fixture has no photo library, so every face is a synthetic portrait painted for it.
This is the page's own honest point: the graph points at the face clusters and can name almost none of them (three of the twenty-nine, in the fixture). It does not need to. Who stands next to whom, how often, and which groups never meet is all there without a single name, and whoever knows the household can put the names back on in a second.
Guessed or given. What you typed in and what the Mac worked out for itself live in the same table. A confidence score is the only thing separating them.
What it recorded
A year of contact. Whom you talked to, when, and not a word of message text (and it does not need any to tell when you are asleep).
Its card on the landing page is the one that gets the reaction, and it contains no inference at all.
A year of who you talked to
A year of contact, per person, split incoming and outgoing, with call counts and a last-seen date, across five messaging apps — and, exactly as the subtitle says, not one word of message text. None of this is content. It is all metadata, and it is enough.
The last line is the one to read twice: the hours it can read as sleep.
iKnow draws that line too, from the same log folded onto a single week:
Weekday against hour of day
Every cell counts the messages and calls that fell in one hour of one weekday, over the year. The working week is in there, and so are the evenings and the late weekend mornings; and so is the pale band from one to seven, every night. Nobody stored a sleep schedule. The quietest six-hour stretch falls out of the timestamps for free.
That is the general shape of this store: almost nothing in it had to be collected, because almost everything in it is a consequence of something that was.
The calendar. The one layer that knows where you will be: the hour, the room, the attendees and the dial-in.
What you withdrew. Withdrawing a fact does not remove it. The row moves to an expired table and keeps its confidence and provenance.
The machinery
The store. The file itself: what it models, and two real rows expanded into everything they touch.
Device fingerprint. Every app installed on the Mac is written down in plain text, next to a token that is an unsalted SHA-256 of that same app identifier, so the page recomputes it in your browser and you can watch it match.
Who else reads it. The graph is the first copy, not the only one. This page maps the views that subscribe to it, the databases they publish into, the tables that repeat the same facts, the streams coming in from hardware paired with the Mac, and the machinery that is set up but has not run yet.
A note on macOS 27. macOS 27 was released on September 14, 2026. After the upgrade to 27.0.1, the graph withdrew most of the rows it had built from Contacts, Calendar, Siri Suggestions, Family Sharing and the app catalog, and rebuilt them over the following day. Some pages of iKnow may not work fully on macOS 27 yet.