One of the most common complaints about AI-generated content is that it all sounds the same: generic, slightly formal and a little too polished to feel authentic. The cause is usually that the tool knows nothing about the organisation it is writing for.
Out of the box, AI tools are trained on enormous amounts of general content. Without specific instructions about how an organisation communicates, they default to an average corporate voice that could belong to anyone. Configuring the tool fixes this far more often than switching to a different one.
What personalising AI involves
Getting AI to sound like an organisation means giving it the right context before it is asked to do anything. That context includes:
- How the organisation communicates, including its tone, its terminology and the phrases it always or never uses.
- Who the audience is, including what they care about, how they talk and what level of technical detail suits them.
- What has already been written, because published material is the best record of an organisation’s actual voice, and blog posts, emails and website copy all show more about it than any written description could.
- Preferences and rules, such as Australian English spelling, phrases to avoid, formatting conventions and anything else that makes the content distinctly the organisation’s.
When these are documented and loaded into the AI tool as instructions, the output shifts noticeably and starts to sound as though it came from the organisation.
Check what goes into the tool
Published material is the safest source of voice. Confidential documents, client information and personal data should stay out of an AI tool until its data terms are understood, including where the information is stored and whether it is used to train the model. Our FAQ on choosing AI tools covers what to check, and rules for staff use sit under Technology Governance.
The AI knowledgebase approach
A structured AI knowledgebase takes this further. Instead of re-explaining the context at the start of every conversation, the organisation builds a set of reference documents that tell its AI tools everything they need to know, once. The same instructions can then be given to every AI tool in use, and each piece of content starts from the right place.
We use this approach for our own content, and it has been tested and documented.
Every draft still needs a person
Even a well-configured tool produces drafts that need editing for accuracy, tone and context. The aim is a stronger first draft that takes less time to finish, with a person still responsible for what is published.
How we help
Setting AI up with an organisation’s context is part of our AI Automation service, which also covers tool selection, a short proof of concept and building the automation around the organisation’s own tools and templates.
Where the need is marketing copy and a content plan, our Communications & Marketing service covers content strategy and delivery, and AI can be part of how that content is produced. For the basics, see our FAQ on content marketing.
