Culture sits in the sentence, not in the away day. A team offsite, a values poster and an annual workshop are moments. What happens every day are thousands of messages: the job ad that deters someone, the feedback that makes someone hesitate, the update that reaches half the room. Culture is what those messages say.
“Culture isn't made at an away day; it's made in the sentence.”
Where culture actually happens
At an away day the mood is good. On Tuesday afternoon someone emails a colleague who has been quiet for weeks. Culture isn't what gets promised at that away day, but what lands in the mailbox afterwards. If you want to know what is really going on, watch the places where people write every day: why policy only lands once it shows up in messages.
- Job ads. Which people apply depends on the words you choose.
- Feedback and conversations. The same trait sounds different when someone is 'passionate' than when someone is 'emotional'.
- Project updates and announcements. Does it reach everyone, or only the people already in the loop?
- Customer emails. That is your brand voice, or it isn't.
What you should measure then
If culture lives in language, it can be followed in language too. Not per person, but per team and over time: does the writing become clearer, more recognisably on brand, more respectful in tone, more inclusive in how it addresses people? The same goes for psychological safety: one survey a year is a photo, language gives you the film.
That can be done anonymously. Above an aggregation threshold you see patterns per team; who wrote what stays out of view. Managers see the direction, not the writer.
Coaching at the moment of writing
A training rarely changes behaviour for good; it is forgotten within a week. What does stick is a hint at the moment you need it. That is why Be-Inc sits inside the programs people already use and explains why it suggests something with every finding, as with tone in work emails. The writer decides, always.
- Explanation. Every suggestion comes with a reason, so people can disagree.
- The human decides. Nothing is auto-corrected.
- Nobody is scored. Reporting only appears above an aggregation threshold.
Norms are set per organisation: what fits one company is not the rule for another. Language policy imposed from above backfires.
What language data does not say
A language pattern is no proof of how someone experiences inclusion. It shows what is written, not what lands. For a picture of culture you combine it with conversation, employee input and other signals. Applied soberly it is a continuous instrument; without context it is a number nobody can do anything with.
Why this is becoming more urgent
AI already writes a growing share of what we send, and that text becomes the input for the next generation of models. Today's tone is tomorrow's default. That is why the defaults matter: the first suggestion, the standard tone, the option that appears first. Whoever sets those quietly sets the norm for thousands of messages at once.
The norm for people and agents
That is why we are building the norm for how organisations communicate. For their people, and for their AI agents. Not a separate tool you open now and then, but the standard everyone writes from: the same brand voice, the same clarity, the same tone and way of addressing people, whether a sentence was written by an employee or by an agent.
So one place holds how an organisation wants to sound, and one place shows whether that line is being met — by people and by agents. What the norm says is up to the organisation.
