Not everything AI produces requires the same level of scrutiny, and recognising the difference is what makes verification practical rather than paralysing.
I work with two tiers. The first tier is high-consequence information: anything a client will act on that affects their safety, their money, their legal standing, or their experience in ways that are difficult to reverse. The second tier is everything else: content, tone, structure, language, presentation.
Tier one requires primary source verification. Every time, without exception. If AI produces a visa requirement, you check it against the relevant embassy or government source. If it produces a health recommendation, you check it against the relevant health authority. If it produces pricing, you check it against the supplier or your booking system. If it produces a property description with specific inclusions, you confirm those inclusions are current. The principle from Module 7 applies: the tool does not carry professional liability. You do.
Tier two requires editorial review. You read it, you evaluate it against your knowledge of the client and the situation, you refine anything that is not quite right, and you confirm it meets your professional standard before it leaves your practice. This is the generation-versus-judgement discipline from Module 1 applied to every piece of output. It takes a fraction of the time you would have spent producing the content from scratch, and it is the step that makes the output yours.
The distinction between the two tiers is not about importance. Both matter. It is about consequence. Tier one errors cause harm that is immediate and sometimes serious. Tier two shortcomings cause embarrassment that is avoidable and usually correctable. Both require your attention. They require different kinds of attention.
