---
title: "What is identity resolution?"
description: "Identity resolution defined: connecting one person's identifiers across channels and sessions so an agent recognizes a returning user."
canonical: https://past.dev/glossary/identity-resolution
last-updated: 2026-09-02
---
# What is identity resolution?

Source: https://past.dev/glossary/identity-resolution

Identity resolution is connecting one person's identifiers across channels and sessions, such as an email address, a chat handle, and a user id. A returning user rarely presents the same identifier twice: the person who emailed last month may open a chat under a display name today. Identity resolution links those identifiers to one profile, so memory written under one channel is recallable from another.

## How identity resolution works

Identity resolution maintains a graph of identifiers: email addresses, chat handles, phone numbers, account ids. Edges assert that two identifiers belong to the same person, each with a recorded reason and a confidence.

- **Deterministic links** join identifiers that share a verified key, such as a chat account registered under the same email address.
- **Probabilistic links** join identifiers on converging evidence, such as the same display name, the same employer, and overlapping meeting attendance.
- **Reversible merges** keep the underlying records intact, so a wrong link can be undone without losing data.

The [identity resolution guide](/guides/identity-resolution) covers match rules, confidence thresholds, and merge reversal in depth; the [testing guide](/guides/testing-entity-resolution) covers verification.

## Identity resolution vs entity resolution

[Entity resolution](/glossary/entity-resolution) is the general operation: linking records that refer to the same real-world entity of any kind, including companies and projects. Identity resolution is the person-focused case, and it adds a concern the general case rarely has: the same person deliberately uses different identifiers in different channels, and each session arrives carrying only one of them.

[Coreference resolution](/glossary/coreference-resolution) is different again: it links mentions inside one text and runs before either of the other two.

## Identity resolution in practice

On May 6, m.okafor@lumenlabs.co signs up for a product. On May 19 a ticket arrives from a personal address whose author writes that she signed up with her work email. On June 2 the same person joins a support call as "Maya O.". Identity resolution links the three identifiers to one profile: the ticket through its stated work email, the call attendee through name and company.

A June query for Maya's open issues now includes the May 19 ticket, even though it arrived under a different address. Without the links, each channel holds a fragment and every new session treats a known user as new. Merges must also be reversible: when a probabilistic link turns out wrong, the profiles need to split cleanly, with each fact returning to the identity its source supports.

## Related concepts

- **[Entity resolution](/glossary/entity-resolution)**: the general record-linking operation identity resolution specializes.
- **[Cross-session memory](/glossary/cross-session-memory)**: what identity resolution makes usable, by connecting a returning user to their history.
- **[Coreference resolution](/glossary/coreference-resolution)**: mention-level linking inside a single document.
- **Evaluation**: the [benchmark methodology](/benchmarks/methodology) describes how agent memory systems are evaluated.

## Frequently asked questions

### Is identity resolution the same as entity resolution?

They overlap. Entity resolution links records for any kind of entity, including organizations. Identity resolution is the person-specific case: it connects one individual's email addresses, handles, and account ids across channels.

### How does identity resolution work across channels?

Systems join identifiers deterministically when a shared key exists, such as the same verified email, and probabilistically when only names and context align. Each link carries a confidence, and low-confidence links stay separate until more evidence arrives.

### Why does identity resolution matter for AI agents?

An agent that cannot connect a chat user to last month's email thread treats a known customer as new. Linking identifiers lets memory written in one channel answer questions asked in another.

## Related

- [Identity resolution guide](https://past.dev/guides/identity-resolution)
- [What is entity resolution?](https://past.dev/glossary/entity-resolution)
- [What is coreference resolution?](https://past.dev/glossary/coreference-resolution)
- [What is cross-session memory?](https://past.dev/glossary/cross-session-memory)
- [Memory API quickstart](https://past.dev/docs/memory-api/quickstart)