Routing Overview

Routing is where support policy becomes executable behavior. Knowledge quality decides what Jardine can say. Routing decides what Jardine should do with a conversation after intent and context are evaluated.

In practical operations, routing is what keeps automation useful instead of risky.

Automatic by Default, Manual When Needed

In the current dashboard, routing starts in an automatic mode. That is intentional. Most teams can get stable behavior quickly without manually building large rule sets.

When your operation needs stricter control, you can switch to manual controls. Manual routing exposes three policy objects:

• tags,
• destinations,
• rules.

Tags define intent categories. Destinations define where escalated conversations are sent. Rules connect tags to destinations with priority and optional conditions.

A good way to approach this is to stay in automatic mode until you have a clear policy reason to move into manual controls. Manual flexibility is powerful, but unnecessary complexity can make behavior harder to predict.

The Three Building Blocks

Tags describe support intent classes in policy terms. Some tags may require human handling, while others can allow broader automated handling.

Destinations are downstream handoff targets. In the current interface, supported destination types include Slack, Discord, generic webhook, and Zapier. Each destination type has its own required config fields.

Rules bind tags to destinations with explicit priority and optional matching conditions. Conditions can include confidence bounds, message length minimums, human-required flags, and message content checks.

When these three blocks are well-designed, escalations feel intentional instead of noisy.

How to Think About Default Behavior

Every organization needs a dependable fallback policy. In routing terms, that means maintaining a usable default classification path and avoiding policy gaps where no clear handling route exists.

In the current routing UI, default behavior is protected so teams do not accidentally remove foundational paths and create brittle escalation outcomes.

This is good guardrail design. It keeps teams from making destructive edits during early experimentation.

Designing Routing for Real-World Stability

The most stable routing setups are usually the simplest.

Start with a small set of tags where each tag corresponds to genuinely different handling behavior. Then add only the destinations you actually use. Finally, create a small number of clear rules with explicit intent.

If your team creates many near-duplicate tags and highly overlapping conditions, routing quickly becomes hard to reason about. When that happens, incidents take longer to diagnose because policy intent is no longer obvious.

A practical question for every new rule is: “What real handling difference does this create?” If the answer is weak, do not add the rule.

When to Use Manual Controls

Manual routing is appropriate when your team needs one or more of the following:

• explicit escalations to different systems by intent,
• strict handling conditions for sensitive categories,
• deterministic policy behavior that can be audited and explained,
• tight integration with downstream workflows in Slack, Discord, webhooks, or Zapier.

If you are still proving first-response quality, stay simple first. If you are scaling into multi-team operational policy, manual controls become more valuable.

Common Routing Failure Patterns

One common failure is condition overfitting. Teams add too many narrow conditions, then wonder why rules rarely match.

Another is taxonomy sprawl. Too many tags with unclear boundaries create unstable classification-to-policy mapping.

A third is destination drift, where old destinations remain configured but no longer align with active support workflows.

A fourth is prioritization confusion. If priorities and conditions are not intentionally designed, rule outcomes can feel random even when the engine is working correctly.

These issues are preventable with periodic routing hygiene reviews.

A Healthy Routing Review Rhythm

A lightweight weekly review keeps routing stable:

1. Inspect escalations that surprised the team.
2. Confirm whether current tags still map to real operational categories.
3. Remove or simplify low-value rules.
4. Verify destination configs are still valid and actively used.
5. Retest sensitive scenarios in validation.

This process keeps policy understandable and reduces incident noise.

Routing and Human Trust

Good routing is not about maximizing escalations or minimizing them. It is about matching the right handling path to the right conversation.

When routing is healthy, customers get faster reliable outcomes, and agents get escalations that truly need judgment. That balance is where trust is built.

If your routing model is still maturing, begin with small, clear policies and evolve carefully. Predictability beats cleverness every time in support operations.

To design robust policy details, continue with Tags and Rules.