Guide
Human-in-the-loop approvals in Rails
Some agent actions should pause for a human. Approvals keep automation fast without giving models unchecked write access.
Updated 2026-07-28 · Rails Agent · Tiny Bubble Company
Direct answer
Human-in-the-loop (HITL) approvals let a Rails AI agent prepare an action—refund, email, ticket update—then wait for a person to approve before execution. Rails Agent treats approvals as part of the production agent lifecycle.
When to require approval
- Money movement or refunds
- External customer messages
- Destructive data changes
- Privilege escalation or account changes
- Ambiguous policy decisions
Design tip
Show the human the proposed tool arguments and rationale from the run. Approvals without context create rubber-stamping.
Frequently asked questions
Can every tool require approval?
Yes in principle, but overusing approvals destroys latency. Reserve HITL for irreversible or high-risk actions.
How does HITL relate to guardrails?
Guardrails constrain what an agent may attempt; approvals gate the attempts that still need a person.
Ship production Rails AI agents
Install rails-agent-stack, open /agents, and go from scaffold to deploy with Playbooks, memory, guardrails, and monitoring included.
Keep reading
Ruby on Rails AI agents →
Category pillar
Build AI agents in Ruby on Rails →
Implementation guide
Getting started docs →
Install and open /agents
Pricing →
Plans and runtime
Budget controls and guardrails →
Policy layer
Support triage use case →
Common HITL domain
Observability: traces and costs →
See every run
Guardrails docs →
Product documentation
Security →
BYOK and workspace controls
Support department hub →
All support solutions
Customer-support agent →
Orders, policy, escalation
Email agent →
Inbound mail with approvals