Unbound

A unified API for personal data across every platform.

What these docs are for

These are the specs we build from — product architecture, data model, and technical implementation details. They serve two audiences:

For developers integrating with Unbound: Everything you need to understand the API contract, data model, and query patterns. Start with the Data Model to understand the six canonical entities, then read the API spec for query and mutation patterns.

For AI agents navigating LifeDB: These specs are optimized for autonomous agent consumption. All personal data normalizes into six entities: Person, PlatformIdentity, Conversation, Message, Event, and Document. Every query requires a QueryScope (tenant ID + data scope filters). The Architecture spec covers the WriteStore/ReadStore split, and Module Interfaces defines the Go interfaces your code calls. For agent-optimized navigation, see AGENTS.md.

Key assumptions

Six canonical entities. All personal data — messages, emails, calendar events, documents — normalizes into Person, PlatformIdentity, Conversation, Message, Event, and Document (plus Attachment as a sub-entity).
Two-tier identity. PlatformIdentity is a person on a specific platform (phone number, email, Slack handle). Person is the resolved real-world identity linking one or more PlatformIdentities.
WriteStore / ReadStore. Writes go to an append-only versioned store. A Projector asynchronously builds denormalized read-optimized views. Reads never touch WriteStore.
QueryScope everywhere. Every Store method requires a QueryScope (tenant ID + optional data scope). This is a compile-time constraint, not a convention — tenant isolation is enforced at the type level.
Provenance on every relationship. Every link between entities carries a confidence tag: platform_reported, exact_match, user_confirmed, heuristic, or ai_suggested. Consumers decide what to trust.

Product Specs

Data Model

The unified schema — six canonical entities, three-tier metadata, and the provenance model. Read more →

Ingestion

Connector architecture, sync modes, bulk import, and reliability guarantees. Read more →

Entity Resolution

Identity linking across platforms, confidence scoring, merge/split semantics. Read more →

API

Read/write operations, filtering, timeline, search, and pagination. Read more →

Technical Specs

Architecture

Module boundaries, data flow, WriteStore/ReadStore split, and deployment topology. Read more →

Data Schema

PostgreSQL schema, WriteStore versioning, ReadStore projections, and indexes. Read more →

API Implementation

GraphQL schema, resolvers, REST endpoints, and authentication middleware. Read more →