Customization

Customization should improve clarity and consistency, not add chaos.

This page covers what you can safely tune in Jardine.

What You Can Customize

You can customize AI profile style, tone instructions, routing behavior, confidence policy, and channel-level generation/send controls.

AI Profiles

In AI Customization, you can choose or create workspace-scoped profiles.

Profiles influence tone, response structure, and how Jardine asks clarifying questions.

Use one profile as the team default to keep customer experience consistent.

Tone and Style Guidelines

Good tone guidance is specific and easy for the team to audit. A practical style direction is: clear and calm language, short answer first, steps only when steps are actually needed, and no filler.

Guardrails You Should Always Keep

Core guardrails should always stay in place: do not invent policy exceptions, do not claim actions that were not completed, do not expose sensitive details without verification, and escalate when policy requires human review.

Channel Controls

Per channel, you can run in Generate only mode or Generate + send mode.

Recommended sequence:

1. Start in generate-only.
2. Review outcomes.
3. Enable send when quality is stable.

Safe Rollout Pattern for Changes

When you change profile or policy:

1. Run validation scenarios first.
2. Canary on limited traffic.
3. Review conversation outcomes.
4. Expand only after stable results.

Small controlled changes are easier to reason about than broad rewrites.