Last week we introduced the hybrid journey. Automate first, hybrid, human first. A model that helps you understand what belongs where.

This week is about governance. Not the glamorous part of AI, but the part that decides whether it actually works. The rules that keep automation useful, predictable and aligned with human judgment across a dealer network.

If Week 1 gave you the language, this week gives you control.

Why governance is harder than it looks

In automotive, nothing is standardised for long. Dealers run different DMS contracts. Processes evolve locally. Advisors carry more operational knowledge than any system. Customers expect both speed and reassurance, often within the same interaction.

Inside that environment, automation becomes fragile if nobody sets the boundaries.

Here is a real scenario from the field. A dealer group automated pre service reminders. Everything worked smoothly until an integration glitch happened. Customers replied with questions that never arrived at a real person. Staff assumed someone else was watching it. Nobody noticed until complaints surfaced. The issue was not the glitch. The issue was silence. No signal. No threshold. No owner.

This is what governance prevents. Not failure itself, but silence around failure.

The three layers that keep automation aligned

Every mature organisation we work with ends up with the same three layers, even if they describe them differently.

First, the journey rules. These define which moments are automated, which are shared between AI and humans and which stay fully human. This is the structure from Week 1 and it becomes the reference for every new idea, feature or process change.

Second, the operational rules. These tell your organisation what happens when automation reaches its limits. They specify the moments when a person must take over, how quickly that should happen, and how the customer experience is recovered if something goes wrong.

Third, the compliance rules. These ensure the customer journey stays within regulatory boundaries (such as GDPR and the EU AI Act) and meets ISO 27001 and 27701 expectations. Many AI projects skip this part or add it only at the end. We design for it from the start, so the workflows you create are safe from both a customer and compliance perspective.

You do not need a manual to explain this. These three layers work together as a quiet operating system for your journey.

The most important rule in this entire series

The Automation Handover Rule.

Every customer facing system has a moment where it should step back and let a human take over. Most organisations never define that moment. This is where things unravel.

The rule is simple. You first define the signals that show a customer needs a person. For example when asking a question that involves price exceptions. Then you define the threshold. This might be a 90 minute limit during business hours for any automated interaction that has not progressed. Then you make sure someone owns that moment. Lead owner, advisor, retention specialist. It does not matter who, as long as it is explicit. And finally you decide how to recover trust when automation has already disappointed a customer. A short call. A personalised message. A corrective next step. Something that shows a human actually noticed.

This rule will not prevent failures. What it does is expose them quickly so customers do not feel abandoned. In large networks, that difference is everything.

What a one page governance guide actually looks like

You do not need a long document. The most effective governance documents should fit easily on one page and contain four elements: the journey definitions, the Handover Rule, the boundaries set by GDPR and ISO, and a clear path for escalation when something breaks.

This page becomes the touchstone for every dealer, every new hire and every process review. When teams disagree, they return to it. When systems evolve, it gets updated. It is the simplest possible anchor for consistency.

If a governance guide cannot be explained in two minutes, nobody will use it. If it can, you have a foundation that scales.

Making governance real across your network

Rolling out governance is not about training slides or policy signoff. It is about embedding new reflexes into your organisation.

You start with dealer managers, because advisors follow their lead (hopefully...). You reinforce the rules inside your tools so the correct actions are guided, not remembered. You discuss breakdowns openly during the first weeks instead of hiding them. And you let the system adjust itself based on what actually happens on the ground.

Governance is not there to slow anyone down. It exists so that automation remains reliable, predictable and aligned with human judgment instead of slipping into inconsistency.

Why this matters

AI will become more capable every month. That does not mean it should run without boundaries. Governance gives you the confidence to scale automation without sacrificing the customer experience. It keeps your teams aligned even when their tools are different. And it protects the human moments that create loyalty.

Week 1 gave you structure. Week 2 gives you clarity. Week 3 will focus on the humans inside this system and how their roles evolve.

Coming next

Governance is the structure. Next week we focus on the people inside it.