SDK Overview — bRRAIn Docs
The bRRAIn SDK embeds AI memory into any application via a Go-first library with HTTP fallbacks for other languages.
SDK Overview
The bRRAIn SDK is an embeddable AI memory engine. It exposes the full bRRAIn platform — workspaces, records, graph retrieval, non-human actors, and domain-adapted Handlers — through a thin client that your application links against.
What you get
- Graph-backed storage — records you write become nodes connected via the POPE ontology
- Semantic + full-text retrieval — sub-second queries enriched with graph context
- Non-human actors — robots, IoT sensors, and automation scripts as first-class session participants
- Domain adapters — swappable Handler LoRA adapters for legal, healthcare, finance, mining, and more
- Zero-trust envelopes — every record encrypted with your derived key
Use cases
- Productized SaaS — add intelligent memory to an existing app without rewriting your database layer
- Embedded AI — pair the SDK with an LLM orchestrator to give agents durable memory
- Real-time systems — ingest telemetry at 1M records/day with graph-based anomaly detection
High-level architecture
Your application code
↓
bRRAIn SDK (Go library / HTTP client)
↓
Handler (local container or hosted)
↓
bRRAIn Core (Vault + Consolidator + Graph)
Key concepts
| Concept | What it is | | --- | --- | | Workspace | A scope for records — typically per-user or per-project | | Record | A schema-agnostic data blob with optional context metadata | | Graph | The POPE-based structure linking records, people, decisions, and events | | Handler | Fine-tuned LLM that classifies, summarizes, and routes | | Session | A logical context for a run of operations (user, service, or agent) |