Canonical set

Five reference flows

These are reusable autonomy patterns, not five demos. PR review and refund approval carry the flagship examples here because they show both the engineer-legible and business-legible sides of the spectrum.

Stacked semantic review

PR review agent

Best for
merge safety
Worker
LLM + tools
Evaluator
A stacked gate that combines deterministic checks with an LLM judge.
Target
output
Risk
medium / technically reversible, operationally costly
Trajectory
HITL -> HOTL
Read the flow
Selective autonomy classifier

Refund approval agent

Best for
selective autonomy
Worker
Rules, classifier, or LLM depending on the implementation.
Evaluator
A separate escalation classifier decides whether the proposed decision is safe to auto-resolve.
Target
output
Risk
medium / partly reversible
Trajectory
HITL -> autonomous for easy cases only
Read the flow
Non-generative classification pipeline

Document routing

Best for
low-risk classifier autonomy
Worker
A routing classifier with no generative model in the loop.
Evaluator
Inline checks and monitoring gate each routing decision.
Target
input and output
Risk
low / reversible
Trajectory
Full autonomy for the in-distribution bulk, with an out-of-distribution escalation path always open.
Read the flow
Irreversible action gate

Transaction / trade execution gate

Best for
irreversible action control
Worker
Upstream automation proposes the transaction or trade.
Evaluator
Deterministic invariants do the heavy lifting.
Target
output and outcome
Risk
catastrophic / irreversible
Trajectory
Deterministic blocking gate always, with human-in-the-loop control for anything that matters.
Read the flow
Evaluator maturity pipeline

Support response agent

Best for
evaluator maturity
Worker
The LLM that drafts the response.
Evaluator
The evaluator matures from an LLM judge to a distilled classifier.
Target
output
Risk
low-medium / reversible
Trajectory
HITL -> HOTL with sampling
Read the flow