interlocute.ai beta
Create a node

Guardrails & Governance

Production AI needs boundaries. Set token quotas, define refusal policies, restrict tool access, and enforce execution limits — all through configuration, not code.

Try Interlocute Read the docs

What are AI guardrails?

Guardrails are policies that define what your AI node can and cannot do. They include token quotas that cap spending, refusal semantics that define when the node should decline a request, access restrictions that limit who can call the node, and execution boundaries that control tool invocation.

Why it matters

Without guardrails, a production AI endpoint is an open-ended cost and liability risk. A single runaway conversation can burn through your token budget. An unrestricted node can be manipulated into producing harmful content. Guardrails make AI deployment commercially and operationally safe.

Configuration, not code

Guardrails are configured through the dashboard or API — no code changes required. Set a daily token quota, define a refusal policy for off-topic requests, restrict access to specific API keys, or limit the number of tool calls per conversation. Changes take effect immediately.

Auditable enforcement

Every guardrail enforcement action is logged: refusals, quota hits, access denials. This creates a clear audit trail that demonstrates your AI operations are governed and compliant with your organizational policies.

Frequently Asked Questions

Guardrails & Governance

What guardrails does Interlocute provide?
Interlocute provides token quotas (daily/monthly spending caps), refusal semantics (conditions under which the node declines a request), access restrictions (API key scoping and domain allowlists), and execution boundaries (tool call limits, rate limiting). All are configurable per node.
How do token quotas work?
You set a maximum token budget for a node over a given period (daily, monthly, or per-key). When the quota is reached, the node returns a clear error indicating the limit has been hit. This prevents unexpected cost spikes and makes spending predictable.
What are refusal semantics?
Refusal semantics define when and how a node should decline to respond. You can configure the node to refuse off-topic requests, requests that violate content policies, or requests from unauthorized origins. Refusals are logged for audit purposes.
Can I restrict which API keys can access a node?
Yes. Each node supports multiple API keys with independent scopes. You can restrict keys to specific capabilities (e.g., read-only, no tool use) and revoke access at any time without affecting other keys or integrations.
Do guardrails require code changes to implement?
No. All guardrails are configured through the dashboard or API. Changes take effect immediately without redeployment. This makes it safe to adjust policies in production as your requirements evolve.
How are guardrail enforcement actions logged?
Every refusal, quota hit, and access denial is logged with the reason, the request that triggered it, and the policy that was enforced. These logs are available through the dashboard and API for auditing and debugging.

Documentation

Guardrails Guide API Reference

Related Features

Commercial Control

Govern usage, enforce quotas, and maintain cost visibility across all your AI nodes. Built for teams that need accountability.

Observability & Logging

Built-in logging, tracing, and token usage visibility for every node call. Understand exactly what your AI is doing and what it costs.

Tool Use & Function Calling

Let your AI nodes call external tools and APIs. Pre-configured function calling with governed execution, built into the runtime.

Ready to build with Guardrails & Governance?

Deploy your node in seconds and start using Guardrails & Governance today.

Try Interlocute Read the docs