Overview

Spectron can be deployed on your own infrastructure so memory and knowledge stay under your control. The self-hosted distribution is a single Rust binary – spectrond – that connects to a SurrealDB instance and an object store. The binary holds no durable state itself (everything important lives in SurrealDB and object storage), so you can scale api and worker replicas horizontally.

When to self-host

Self-hosting is appropriate when:

• Your data must not leave your infrastructure for compliance or privacy reasons.

• You need to customise extraction models, use self-hosted LLMs, or configure the pipeline in ways not supported by the hosted offering.

• You are building on top of an existing SurrealDB deployment and want tight integration.

• You require air-gapped or private-cloud operation.

The hosted offering at cloud.surrealdb.com handles infrastructure, upgrades, and scaling automatically. Self-hosting is an operational commitment – you are responsible for availability, backups, and upgrades.

Deployment options

All three options use the same binary and configuration model. You can start with Docker Compose locally and deploy to Kubernetes in production without any changes to Spectron's configuration – only the infrastructure wrapper changes.

What you need

Every Spectron deployment requires:

• SurrealDB – ws:// or wss:// endpoint. Spectron uses SurrealDB for all durable state: sessions, turns, extracted entities, knowledge nodes, API keys, and decision traces.

• Object store – an S3-compatible bucket, Google Cloud Storage bucket, Azure Blob container, or local filesystem path. Spectron stores the raw bytes of uploaded documents here.

• LLM provider keys – at least one of OPENAI_API_KEY or ANTHROPIC_API_KEY, depending on which models you configure. The extraction, reflection, and response model stages each call an LLM.

See Architecture overview for how these components fit together.

Sections in this guide