Throughout the last weeks, I have had some challenging conversations – or communication triggers you might say. Some came up in the dialogue with our Ägile Ässets users, some within a legal context, some other in cooperation.
Many of those instance in the meanwhile have resolved positively or very positively (in case pulling the switch from “critical” to enthusiastic) in the meanwhile. As an entrepreneur for whom customer happiness is paramount as well as a communication expert who appreciates an elegant solution that works for both sides, I must admit, that I am pretty happy and also a bit proud of that. So, now you’re all already on the hook I hope
how I managed to resolve that with just one super clever prompt that made Claude or GPT give me the perfect answers.
Hehe.
This is where the plot twist happens:
It did not.
Quite the contrary – at some point, it would even have put these situations at risk.
Let’s look into how did I get that done anyway – where and how did I use the AI – and what made it work?
What is the scalable take away?
For one thing, human communication is subtle.
The same words can easily mean very different things if you give them a bit of a different context.
They can mean totally different things depending on the perspective and cultural backdrop of a person.
These things are subtle and the only way to get a real feel is to…
dig deeper.
Feel into the message.
What are the assumptions this person has made about the world – and our connection – so far?
What does that person want from me – like: really?
Under the maybe polite but also bitchy surface?
What makes them insecure -and what kind of security or safety do they want from me?
One might simply want to be heard, another might want to make sure to cover against what has been sold to them as overall life risk…. They might be used to a normality that looks different than mine and also directs their expectations….
I have tried to make the AI work more subtle on this, giving it prompts such as “consider this from the perspective of… be sure to differentiate between things we can and should teflon vs. such that really need treatment” etc. Maybe you can make it work, but me even with the longest possible and a pretty broad experience in prompting (we use all the big tools since they exist) plus my 20yrs of experience in briefing people (from my agency / consulting background)
could not really get them to get to an answer that felt “Just right”.
This here is important though:
“Just right” – might be just 5% difference (though here it was far more than that), but it makes ALL, as in 100% a difference in terms of how the other person will perceive the message and react.
Do not let yourself be mistaken, and think “this does not really matter” or is overdone perfectionism.
If your relations with your clients, work partners, etc. matter, these 5% will matter.
It will take more effort and time shaping it – but… and this is great news… in my experience in 7 of 10, maybe 8 of 10 cases people will literally SMELL the difference.
Like this, the extra effort you put in, will pay large dividends either neutralizing conflict before it even exists or turning it around or creating real fans.
Part of the problem seems to stem from the AI tools not being able to really count in weight and nuance.
Weight for me would be the fact that half a phrase or even one word might throw a totally different light or angle on the whole copy. Like this, this word or expression would have to be much higher prioritized and the message reflected within that context.
Nuance is the more subtle little brother, it means that the choice of words, where it is wordy or curt etc. all carry meaning. This part is a bit like Scrabble, it says that words do have different ‘value’ within a certain contect – and in contrast to Scrabble, this means that the value of one word of phrase might change.
That is what makes communication, literature so timelessly compelling – however, you cannot simply reduce them to the core meaning because then you’re missing the REAL meaning – and how can your answer then make the other person feel heard.
If you clearly did not really read or listen closely?
So, how do you go ahead to make it work?
It might make sense to let a good AI help you dissect what the messages / communication is about so far. It might help you open up to blind spots, nuances you have not really noticed.
Maybe give that another loop, ask for different versions (pretend I am very authoritative vs. very forgiving vs. X), then read through those. This is not supposed for you to take it over, but for you to get a sense of the differences and what feels adequate.
Then, at best, let it sit.
On the next day, come back, read through the original message at first, then jot down with pen and paper what now comes to mind.
Revisit your intention and stance: this could be something “I want to be helpful and assuring, at the same time be transparent about our limits”. “I want to convey on eye level, that A is not ok given that B, that what I wish for is X or maybe Z. I am open to talk.” In most cases, it is good to give it some time between first reading and reaction. More often than not if a message is not that easy to answer or overall send, there will be some part that simply triggers us. For me that might be an Erbsenzähler mindset on the other side, or perceived injustice… Acknowledging the trigger from my POV is the key to being able to let it go and go beyond – into the connection and what needs to be clarified or how to strengthen the relationship.
Now write the new version. maybe check if there were some nuggets from the AI you can reuse (for me, it is often stuff that is matter of fact indeed – as in questions around GDPR – Claude is better in summing that up in a no BS way). do not take anything copy/paste without a critical eye – from my own experience I can tell you people working with AI a lot will SEE the AI touch.
Part of the trick is to use it more in the sense of the creative technique of “diverging” vs. “converging” – in converging you collect perspective, elements, etc – in converging you deliberately throw away, ruthlessly dissect what’s to be kept and what can go. If that is only 1 of 10, that is just fine.
Should you still use AI for it even if you skip it all in the end?
Yes, I would still keep that up for the simple reason that it gives my reflection process distance. It is easier for me sometimes to develop a clear understanding of what I do not want to convey – simply by contrast. It makes me use the Viktor Frankl insight: “Inbetween stimulus and response, there is a space – and within that lies our freedom (of choice).”
So, yes, each of those message was more work, at first.
But each helped remove social friction rather than create it.
With each, again, after over 20 years I learned a LOT or was reminded that taking care of detail is a valuable invest.
It’s like the difference between just slapping random whitebread together with supermarket bacon vs. cooking with love and great ingredients.
If those relationships matter (and they typically will or COULD), I should be willing to go that tiny extra (semi-)mile.
Indeed, the learnings itself seem to be a reward because it teaches me things about communication that are still new to me even though I have studied and worked in that field for over 20 years.