Itjoe
ENNL

Copilot for communications

Copilot for communications: writing from your own sources.

For communications, marketing and policy teams producing Dutch copy all day. The gain isn't in writing faster, but in the first draft and what you do with it.

The work shifts, it doesn't vanish

01 · What changes

Writing a first version takes less time. Judging it takes more. Skip the second and sooner or later you publish something that isn't true.

For a communications team that's an uncomfortable trade. The first draft was always the moment you thought about what you actually wanted to say; now something fluent appears immediately and you respond to it instead of thinking for yourself. That's the main trap, and no amount of technology solves it.

What does work: bring Copilot in only after you've decided the core. Write three lines yourself about what has to be in it and why, then let it flesh that out. The result is better and, more importantly, it stays your piece.

The second shift is tone of voice. Without instruction Copilot picks the safest Dutch: formal address, long sentences, nominal style. For a letter to a citizen that's sometimes right, and for an intranet post almost always wrong. One shared prompt rule with your tone guidelines fixes that across the organisation, fifty people each enforcing their own tone does not.

The limit

An invented figure comes out just as fluently as a correct one. There's no hesitation in the tone to tip you off. Anything going public that contains a fact gets checked at the source, however well it reads.

Where it saves time immediately

02 · Concretely

Word

Reducing a long document

Turning a twenty-page policy document into a half-page summary, or into a plain-language version. Point at the document rather than pasting it: that way it keeps the structure.

Outlook

Variants of one message

Ten subject lines, three lengths of the same email, or the same message for a different audience. Generating variants is where the model is unambiguously faster than you.

PowerPoint

From document to deck

Turning an existing document into slides. The structure usually comes out usable; the design always needs manual work. Expect a good starting point, not a finished product.

What to arrange up front

03 · Agreements

01

Tone rules in one place

Write your tone of voice out as three to five prompt lines and put them in the prompt gallery. Everyone pastes the same block, so everyone gets the same tone back.

02

Publishing rules on one page

What goes out unchecked, what needs a second reader, and what we always write ourselves. Shorter than a page and it isn't policy; longer and nobody reads it.

03

Agree what you never delegate

Crisis communications, personal messages and anything with legal weight. Not because it can't be done, but because the cost of one mistake there outweighs any time saved.

Questions from communications teams

04 · Frequently asked

Can Copilot hold on to our tone of voice?

Within one conversation, yes, if you supply the rules or point to a document containing them. Across conversations, no: Copilot remembers nothing. In practice a fixed block of text everyone pastes in works better than explaining it each time.

Does this make my writing job redundant?

No, but it shifts. Writing the first draft takes less time; judging whether it's right takes more. Someone who can only write becomes less valuable. Someone who can spot what's wrong with a smoothly written text becomes more valuable.

May I use Copilot for text that goes public?

That's a policy question, not a technical one. Make agreements and put them on a single page: what goes out unchecked, what needs a second pair of eyes, and what we always write ourselves. Without those agreements everyone does something different and you find out when it goes wrong.

Does Copilot notice when a text is wrong?

No, and that's the hardest part. An invented figure or a misexplained term comes out just as fluently as a correct fact. There's no signal in the tone that gives it away. That's why there's a whole module on judgement in the training.

What does it concretely deliver for communications teams?

Mostly time on the tedious parts: reducing a long document to a summary, rewriting text to a different reading level, or generating ten subject-line variants. The creative part, deciding what you want to say, barely changes.

Read on

First measure what your team already knows.

A communications team is rarely at one level. The baseline shows where the gaps are, and only then do we decide whether a workshop still adds anything.