I once sat in a forty-minute meeting that ended with the phrase "great, I think we're aligned." Everyone nodded. Everyone left. Two of us went and built different things.
We were not aligned. We had agreed on a sentence. The sentence had been carefully constructed, by several intelligent adults, to be true no matter what happened next.
That is the thing about workplace communication. Most of it is technically English.
A short dictionary
Here is a starter set. You will recognise all of them, because you have said at least four of them this month.
| What was said | What it meant |
|---|---|
| Let's take this offline | I am not doing this in front of witnesses |
| I have some concerns | This is a bad idea and I have decided to be diplomatic about it exactly once |
| Just to play devil's advocate | I hold this view sincerely and want deniability |
| Let's circle back on this | Never |
| Can we double-click on that? | I stopped listening two slides ago and need to look engaged |
| Happy to discuss | Unhappy to discuss |
| As per my last email | You did not read my last email |
| Quick question | Forty-five minutes |
| Let's socialise the idea first | I want it to be someone else's idea by the time it is decided |
| I'll take a look and get back to you | The look is optional |
| Interesting | No |
| Very interesting | Absolutely not |
| Let's park it | It is dead, but a burial would require a decision |
| We should probably loop in Legal | I would like this to stop moving |
| No rush on this | Today |
| Does that make sense? | Please agree so this ends |
| Thoughts? | Approve it |
| Bandwidth | Time, but expensive-sounding |
| Let's align on the ask | Nobody in this meeting knows what we want |
| Great question | I do not have an answer and needed a second |
The funny part is how fluent we all are. Nobody is confused in the moment. Everyone decodes it live, in real time, at full speed, and then goes back to their desk with a slightly different translation from the person sitting next to them.
The vagueness is not politeness
Here is where it stops being a joke.
We tell ourselves this language is about being nice. Softening things. Keeping the room comfortable. Some of it genuinely is, and I am not campaigning for a workplace where people say "your plan is bad" and walk out.
But most of it is doing something else. Every hedge moves risk from the speaker to the listener.
"I have some concerns" is not a softer version of "do not do this." It is a version that cannot be quoted back at you in six months. If the project succeeds, you were supportive. If it fails, you raised concerns. You have taken a position that pays out either way, and the cost of that option is paid by whoever had to guess what you actually meant.
"Let's circle back" is not scheduling. It is a way of declining without owning the decline. "Let's park it" kills the idea while leaving no fingerprints. "Let's socialise it first" spreads authorship so thin that no single person can be blamed for it later.
This is not rudeness dressed up as courtesy. It is accountability being quietly redistributed, one pleasant phrase at a time. And it always flows the same direction: away from the person with the most authority, towards the person who has to go and actually do something.
The tell
There is one reliable test, and it takes ten seconds.
At the end of any conversation, ask someone to repeat back what was decided, who owns it, and by when. Not "are we aligned." Ask for the sentence.
If three people give three answers, you did not have a meeting. You had a group hedging exercise with catering.
I do this now on anything that matters, and it is mildly embarrassing how often it fails. Not because people are stupid or lazy, but because the conversation was never optimised to produce a decision. It was optimised to produce comfort. Those are different products and only one of them ships.
Agents do not do subtext
There is a reason this has been on my mind lately.
I run a fair amount of my work through AI agents now, and an agent is the most literal colleague you will ever have. It has never once inferred what I really wanted. It does exactly what the words say, at speed, with total confidence.
Tell an agent to "clean this up a bit" and you will get a lesson in what "a bit" means to something with no fear of consequences.
The first few weeks of that are humbling. You discover that half of your instructions were never instructions. They were vibes, delivered in the reasonable expectation that a human would fill in the missing eighty percent from context, tone, and a shared understanding of what the boss is like when annoyed.
Agents will not do that. So you learn to write the actual thing. What outcome, what constraints, what done looks like, what to do when it is ambiguous.
And then you notice the obvious. Every instruction that had to be rewritten for the agent was also unclear for the humans. They were just better at hiding it. A person who does not understand your request will produce something plausible and adjacent, exactly like a model will, except they will do it over three weeks and then be upset about the feedback.
Working with agents did not teach me a new communication skill. It removed my ability to pretend I already had one.
For the person doing the work
If you are the one receiving these phrases, the practical move is to convert them into something falsifiable, out loud, in the room.
"So the decision is X, I own it, and I'll have it by Thursday. Anything I've got wrong?"
It feels blunt the first few times. Do it anyway. Worst case, someone corrects you, which is the entire point. Best case, you have just created a record of a real decision in an environment that was busy avoiding one.
And when you get "some concerns", ask the follow-up: what would change your mind? A real concern has an answer to that question. A positioning move does not, and you will hear the difference immediately.
For the person deciding
If you lead, the uncomfortable version is this. Your team is fluent in your hedges, and they are spending real energy decoding you.
Every time you say "interesting" and mean no, someone goes away and does another week of work on a thing you have already killed. You did not save their feelings. You spent their time to protect your own comfort, and you did it in a way that made it hard to trace back to you.
Say the thing. "No, and here's why" costs you thirty awkward seconds and saves them a fortnight. That trade is not close.
Ambiguity is a luxury good. The more senior you are, the more you can afford it, and the more expensive it is for everyone below you. Which is precisely why you should stop buying it.
The takeaway
Corporate language is not meaningless. It means something very specific, which is: I would like the upside of having spoken without the downside of having said anything.
The dictionary is funny because we all speak it. It is worth taking seriously because every entry in it is a small transfer of risk from someone who has power to someone who has a deadline.
You do not fix that with a communications workshop. You fix it by being the person in the room who says the actual sentence, names the owner, and puts a date on it. It is not a hard skill. It is just an uncomfortable one, and almost nobody does it, which is exactly why it is worth doing.
Anyway. Great question. Let's take this offline.
Written by
Martin Dimoski
Senior R&D Executive & AI Systems Builder