ads-route

What problem does it solve?

ads-route prevents costly, high-risk decisions from being made with expensive AI calls by routing each decision through a tiered cost-and-risk policy.

Core Features & Use Cases

• Tiered cost firewall: Selects whether to use Brain A (cheap), Brain B (reviewer), or require human approval based on domain, intent, and payload tier.
• Risk scoring and policy matrix: Uses domain-specific risk scoring to map outcomes to ALLOW_BRAIN_A, REQUIRE_BRAIN_B, REQUIRE_HUMAN_REVIEW, or BLOCK_ALWAYS.
• Guardrails for Tier 0 safety: Blocks actions that lack a required “ADR mother,” reducing uncontrolled changes to token/MCP schemas and production-sensitive areas.

Use case example: before triggering a Brain B review or a production deploy, the system routes the request according to whether it touches DESIGN vs PRODUTO and how “CRIT” the risk score is.

Quick Start

Ask an AI agent to route a planned ads-related decision by providing domain, intent, and payload so it returns the correct ALLOW_BRAIN_A, REQUIRE_BRAIN_B, REQUIRE_HUMAN_REVIEW, or BLOCK_ALWAYS outcome.