# Set the guardrails once, not for every change

> Each organisation sets its own policies and risk tiers in Substrait. Changes inside an approved boundary ship freely; only a change that crosses it goes back for approval.

URL: https://substrait.build/capabilities/policies-and-approvals/  
Last updated: 2026-11-06

On Substrait an approval sets a boundary, not a frozen snapshot. Builders change their tool freely inside the boundary, and only a change that crosses it, such as new data, writing instead of reading, or facing customers, goes back for approval.

## Risk tiers

Substrait places every tool in a risk tier by what it can affect: read-only, takes actions, or faces customers.
The tier sets the approval bar and the rules the tool runs under, for its whole life.
Your organisation names its own tiers and severity levels, and Substrait uses your names.

## Approval routing

A change that satisfies policy is approved automatically.
A change the policy flags is routed to the right approvers, with deadlines and escalation.
Approvals can need one person, any of several, or all of several.
Nobody approves their own change.

## The boundary, approved once

An approval defines limits: which data, read or write, customer-facing or not.
Inside those limits, builders keep changing the tool with no new review.
A change that breaks a limit goes back through approval.
That is how Substrait keeps building both governed and fast.

## Sensitive data sets its own rule

For genuinely sensitive fields, your organisation can set a rule on which tools may ship at all if they use them, on top of who may see the data.

## Questions and answers

### Does every change need a human approver?

No. On Substrait a change that satisfies the organisation's policy is approved automatically. A human approves only when the policy calls for it, typically for higher risk tiers, sensitive data or customer-facing tools.

### Who writes the policies?

Your organisation does. Substrait provides the framework and sensible defaults, and your people decide the risk tiers, the approvers and the rules.

## Related

- [A security check before every release reaches production](https://substrait.build/capabilities/security-check/): Substrait scans every release for leaked secrets, weak code, vulnerable dependencies, container issues and problems in the running tool. A tool reaches production only through a promotion that clears the check.
- [See every tool your people built](https://substrait.build/capabilities/tool-registry/): Substrait registers every tool, agent, Slate, skill and knowledge base with an owner, a purpose, a risk tier and a version, and gives every one a switch that stops it.
- [How to govern the apps and tools people build with AI](https://substrait.build/problems/govern-tools-built-with-ai/): People across a company now build apps, agents and Slates with AI. Substrait puts every one in one registry, with an owner, a risk tier, a security check and an audit trail.
- [The communications centre: one pre-wired exchange for the whole organisation](https://substrait.build/capabilities/communications/): The communications centre is one central, pre-wired exchange that connects every alert, approval, notification and message across the organisation. Every tool built on Substrait calls it, so no tool builds its own. The company hooks in its own Slack, Google Chat, Teams, email or WhatsApp, and each person chooses the channel.