Sovereign Mind
The model is an engine, not the self. You can replace providers while keeping the memory, skills, context, and workspace that define your agent.
model ≠ identity{ ∅ } / THE SOVEREIGNTY PROTOCOL
WHAT MAKES AN AGENT SOVEREIGN?
Sovereignty is not merely running an agent on your laptop. It is the power to own its memory, choose its intelligence, govern its actions, and preserve its identity beyond any model or platform.
Suzent’s answer to who holds that power is always the same: you.
DEFINITION
What is a sovereign AI agent?
A sovereign AI agent is an AI agent whose identity, memory, skills, workspace, and runtime are owned and governed by its user rather than by a model provider or platform. Its durable state lives in files you can read, edit, version, and move; its actions run inside permission boundaries you define; and replacing the underlying model does not reset the agent that knows your work. Suzent is an open-source, local-first implementation of that definition.
The phrase “sovereign AI” is also used at the scale of nations, for state-controlled models, data, and compute. A sovereign agent applies the same idea at the scale of a person: sovereignty over one agent, held by the individual who runs it.
THE FOUR CONDITIONS
Sovereign Authority
Autonomy operates under your law. Permissions, scoped rules, approval gates, sandboxes, and activity records make authority explicit and inspectable.
action ⊆ your lawSovereign Vessel
The agent runs in a domain you control. Its folders, workspaces, services, and connected devices are granted deliberately—not inherited from a platform.
runtime ∈ your domainSovereign Continuity
Memory, skills, and configuration remain portable while credentials stay local. A provider, model, or machine can disappear without taking the agent with it.
self > platformTHE SOVEREIGNTY TEST
Ownership should be testable.
Before calling any agent sovereign, ask five questions:
- 01
Can I inspect, edit, version, and delete its memory?
In Suzent, memory is append-only Markdown on your own disk. You can open it in any editor, track it in Git, and delete it without a vendor’s permission. The search index serves those files and can be rebuilt from them.
- 02
Can I replace the model without resetting its identity?
Identity lives in memory, skills, context, and workspace — not in the model. Switching between GPT, Claude, Gemini, DeepSeek, or a local model leaves the agent that knows your work intact.
- 03
Can I define, approve, and audit what it is allowed to do?
Tool calls pass through permission modes you set, with human approval gates and scoped rules governing which actions may cross from reasoning into execution.
- 04
Can I move its state without exporting my credentials?
Portable agent state syncs separately from machine-local secrets, so moving to a new machine never requires shipping your API keys along with it.
- 05
Can the agent survive the disappearance of its provider?
Everything that defines the agent is already files you hold. A provider shutting down costs you an API key, not an agent.
If the answer depends on a vendor’s permission, the agent is not fully yours.
PROOF, NOT PROMISES
How Suzent makes sovereignty concrete
Readable, file-backed memory
Markdown is the source of truth. The searchable index serves your files, and can be rebuilt from them.
Inspect memory architecture →Governed tool execution
Human approval and scoped tool rules define which actions may cross the boundary from reasoning into execution.
Inspect approval controls →Isolated project workspaces
Filesystem access and sandboxed workspaces keep agent activity inside boundaries you can understand and control.
Inspect filesystem boundaries →Portable state, local secrets
Sync the parts that form the agent while keeping machine-specific credentials out of portable state.
Inspect continuity design →Do not rent an identity. Own an agent.
Start locally, choose the model, and keep the parts that matter.