1 Where AI saves real time
AI is at its best on work that is useful to speed up and easy for a person to check afterwards:
- turning rough notes into a first draft that someone then rewrites;
- suggesting several headlines or angles to choose from;
- summarising long documents, such as customer feedback or meeting notes;
- adapting an approved text into shorter versions for different channels;
- producing visual concepts and mood boards before a real shoot or design.
In each case the AI produces options or a starting point. A person still decides what is kept.
2 Where it needs a person beside it
AI tools write fluently even when they are wrong. They can invent facts, figures and sources that look convincing. Anything that states a fact about your business, your prices, a law or a result must be checked by someone who knows the answer.
Tone needs the same care. A text can be correct and still sound unlike your brand, or miss how your customers speak. Translation is a common example: a machine translation may be accurate word by word and still read as foreign to a native speaker, so a native speaker should review it.
3 Where it should not be used
- Inventing testimonials, reviews, case studies or statistics. This misleads customers, whether a person or a tool wrote them.
- Making promises on your behalf, such as guaranteed results, prices or delivery dates.
- Processing customers’ personal data in tools whose data handling you have not checked.
- Publishing anything that nobody has read in full.
4 Being open about AI
When an image or a piece of content is generated with AI, saying so is simple and builds trust. A short note such as “image generated with AI” costs nothing, and it avoids the impression that a staged product shot or an invented scene is real.
This is also the rule we follow at ZENNTAC: AI may help with research, drafting and visuals, and a person reviews and edits everything before it is published.
5 A sensible way to start
- Pick one repetitive task that takes time each week, such as drafting social posts from an approved message.
- Write down what a good result looks like, so the output can be checked against it.
- Decide who reviews the output and what they check: facts, tone and anything legal.
- Try it for a few weeks, then keep it, adjust it or drop it.
In short: let AI speed up drafts and options; keep facts, promises and final decisions with people.