Layer 04
Governed Agents
Act without uncontrolled authority.
An agent is useful because it can do things. It is only safe to deploy when you can say exactly what it is allowed to do. Governed agents act inside an identity, a set of permissions and a set of policies, and everything they do is recorded.
What governed agents can do
Agents on the platform can:
- Reason
- Retrieve
- Analyse
- Use tools
- Communicate
- Recommend
- Decide
- Execute
Every agent works within bounds
Whatever an agent can do, it does inside:
- Identity
- Permissions
- Policies
- Data boundaries
- Tool access
- Risk thresholds
- Approval requirements
- Operating constraints
Build, govern, observe
Build
Studio is a no-code environment for agents, chatflows and automations. BuildX applies multi-agent orchestration to software development.
Read more about BuildGovern
Nexus enforces policy in-flight: it can block what an agent should not do, and traces every agent and connection.
Read more about GovernObserve
Every action is captured, so you can see what agents did, why, and at what cost.
Controlled autonomy, not manual versus autonomous
Autonomy is a dial you set per agent and per task. The platform is designed to let you raise it as trust is earned.
- 01
L1 Assist
AI recommends.
- 02
L2 Approve
A human approves.
- 03
L3 Supervise
AI acts within limits and is monitored.
- 04
L4 Autonomous
AI works independently within strict policy and risk bounds.
- 05
L5 Adaptive
AI improves within controlled boundaries.
What “governed” looks like in practice
Take an AI Procurement Agent. It can read approved supplier information, compare pricing and create draft purchase orders. It cannot make payments or approve its own high-value transaction, and purchases above a threshold need a person. The full example, with what it records, is on the governance page.
Where this sits in the closed intelligence loop
Control, then intelligence, then agency, then execution, then outcomes. Outcomes and their evidence flow back into data and context, so the loop closes.
- 01 · Layer 01ControlDecide where and how AI runs.
- 02 · Layers 02 and 03IntelligenceTurn data and knowledge into useful intelligence.
- 03 · Layer 04AgencyLet AI act within defined authority.(this page)
- 04 · Layer 05ExecutionEmbed that action in how the organisation operates.
- 05 · Layer 06OutcomesMeasure business results, and feed evidence back into data and context.
This layer is the Agency stage: AI acting, within authority you define.
Where Swfte products sit in this layer
Product names are kept. Each is an entry point into the platform.
Studio
Build agents, chatflows and automations without code.
Nexus
Captures every agent action, enforces policy in-flight and traces every agent and connection.
BuildX
Multi-agent orchestration that develops software from natural-language descriptions.
Connect
The model gateway and routing layer agents call: one API, fallbacks, usage caps and an audit trail.
What this means for your team
- Chief AI Officer
- Move from AI experiments to governed production AI.
- CISO
- Give AI identity, permissions, policies and auditable controls.
- CTO
- Deploy models, agents and workflows on infrastructure you control.
Frequently asked questions
What is a governed AI agent?
An agent that acts within a defined identity, permissions, policies, data boundaries, tool access, risk thresholds, approval requirements and operating constraints, and whose actions are recorded.
How is a governed agent different from an agent with guardrails?
Guardrails ask whether an output should be blocked. Governance also asks who is acting, what they are allowed to do, under which policy, with which data and model, at what risk and with what oversight, and whether you can prove it.
Can an agent approve its own actions?
Not where policy forbids it. Approval requirements, such as a person approving purchases above a threshold, are part of the agent’s definition, and an agent cannot approve its own high-value transaction.
What do you mean by controlled autonomy?
Five levels, from AI that recommends to AI that improves within controlled boundaries. You choose the level per agent and task, and policy limits what each level may do. See the autonomy page for detail.
Which Swfte products build and govern agents?
Studio builds them, Nexus governs and observes them, and BuildX orchestrates multi-agent software development. Cortex puts engineers’ own AI agents under one policy and one audit trail.
Across every layer
These ideas apply to every layer of the platform.
Controlled autonomy
Five levels, from AI that recommends to AI that adapts within limits.
Trust Profile
The record that describes what each AI system is and may do.
AI sovereignty
Seven kinds of control over your AI estate.
AI governance
Governance that runs inside AI, not beside it.
Swfte Intelligence Platform
Analyse and visualise your data, usage, agents and outcomes, then build from what you see.
Company brain
One governed place for everything your organisation knows, for every product to read from.
Custom models
Models adapted on your own data, evaluated before release and deployed under your control.
How Swfte builds with Cortex
An honest first-party account of how we build, review and approve work, labelled by status.
Governed agents
Identity, permissions, policy, approvals and audit for agents, with ten templates.
Compliance-approved workflows
Workflows with approval gates and evidence built in, with twelve templates.
Build on Governed Agents with Swfte
Start with one entry point. Add intelligence, agents, workflows and infrastructure as you prove value.