Skip to main content
Sage is a decision model: send content and a question, get back a decision your code can act on. Use it to route, approve, block, score, rank, tag or escalate. No training needed. When Sage isn’t sure, the answer is null, so your code knows to hand off to a human.

​
Decision kinds

Yes/No

Yes, no, or null.

Choice

Pick one option, or null.

Scale

Score against your rubric.

Sort

Rank items by a criterion.

Tags

A verdict per label.

Quickstart

Make your first call.

​
Where it fits

Agents

Routing, branching and escalation in agent workflows.

Operations

Triage and route work across DevOps, SecOps and support.

Guardrails

Check tool calls and responses before they run.

Moderation

Tag, score and rank content in data pipelines.

Risk and fraud

Flag risky transactions, trades and agent actions such as refunds.