The Pre-Agentic Audit: What To Fix Before You Let AI Handle Your CX
Dubai has given its private sector a two-year clock.
In May 2026, Sheikh Hamdan bin Mohammed launched an initiative - run through the Dubai Chamber of Commerce, covering roughly 295,000 affiliated companies - to move the entire private sector toward agentic AI.
Training tracks, incubators, dedicated funds. The direction of travel is clear even where the enforcement detail isn't published yet.
Most of the conversation since has been about which tools to adopt and how fast.
Almost none of it has been about what those tools will actually be automating.
The question nobody's asking
An agentic AI system doesn't fix your customer experience.
It runs it - faster, and without a human noticing when something's gone wrong.
If your current process quietly loses leads in the first 48 hours, an AI agent running that same process will lose them faster and more consistently.
If your onboarding has a drop-off point nobody's ever measured, automating it doesn't remove the drop-off - it just means fewer people are around to notice it happening.
Agentic AI amplifies whatever process it's given. That's the whole point of it. It's also the risk nobody's pricing in.
Why this matters before you automate, not after
Once an agent is live and handling real customer interactions, unwinding a bad process is a lot more expensive than fixing it would have been on day one.
You're no longer changing a workflow - you're retraining a system, re-auditing outcomes, and explaining to customers why something changed twice.
The businesses that get real value out of agentic AI over the next two years won't be the ones who moved fastest. They'll be the ones who knew exactly what they were handing over before they handed it over.
That's a diagnostic question, not a technology question. It's "where does our current customer journey actually leak, and by how much" - answered properly, with real numbers, before any of it gets automated.
What a pre-agentic audit actually checks
Three things, specifically:
Where the leaks already are.
Every business has at least one stage where interested customers quietly drop off - a follow-up that doesn't happen consistently, an onboarding step that loses people, a review response that never goes out. Automating around an unmeasured leak just embeds it.
What "good" looks like in numbers.
An AI agent needs a target, not a vague instruction to "handle it well." Without a measured baseline, there's no way to know whether the agent is actually performing better than the human process it replaced or just doing the same thing faster.
Which parts of the journey should stay human.
Not everything should be handed to an agent. Knowing which touch-points depend on human judgement and which are pure process, is itself a diagnostic decision, not a technology one.
The practical starting point
You don't need to wait for a vendor conversation to start this. A CX diagnostic done properly, measuring where the leaks are and what they're costing, before anything gets automated is the same work either way.
The only difference is whether you do it before you hand your customer journey to an AI agent, or after it's already run for six months on a process nobody checked.
Given the timeline Dubai's set, that's not a decision with much room to be casual about.
Where to start
If you're not sure whether your own customer journey has a leak worth knowing about before you automate anything, a Fix-It Call is the place to start.
A free, no-obligation conversation to talk through where the risk actually sits in your specific process. No pitch, just a second pair of eyes before you hand anything over to an agent.
Comments