What I Didn’t Know I Wasn’t Saying

18/07/2025

How AI helped me to identify (and have) the conversations I wasn’t having.

There is a particular intimacy in the way we message each other. Often quietly, sometimes urgently, we move through our days exchanging fragments of thought and feeling in digital form. Which got me thinking – I wanted to know What I didn’t know I wasn’t saying – if that even makes sense. [it doesn’t – ed]

What am I sitting on, essentially, instead of articulating.

A quick reassurance here, a well-timed joke, a carefully phrased update that conceals more than it says. We write as ourselves, or at least, as a version of ourselves we believe will be understood.

When we first invited readers to upload their WhatsApp chats into an AI tool and ask what they revealed about their relationships, we didn’t know quite how far the invitation would travel. What began as an experiment in emotional curiosity quickly became something else – something raw, occasionally confronting, and for many, unexpectedly moving. 

We heard from people who recognised communication patterns they’d never named, who saw affection where they feared distance, or imbalance where they had assumed ease. For some, it affirmed connection. For others, it pointed – gently but unmistakably – towards conversations that hadn’t yet been had.

This piece is a continuation of that experiment, and also a deepening of it. If the first article held up a mirror to our relational style, this one asks us to pause in the silences and subtleties. To explore the gaps in our communication – the things we edit out, hold back, soften, or avoid altogether. Not to judge, but to notice. And in noticing, to understand ourselves a little better.

Once again, there’s a simple practice you can try. Because what’s all the talk without curiosity and fun, eh?

 

The “What I Didn’t Say” Experiment (in Two Parts)

We don’t always realise how much we hold back in our communication. Not just in what we avoid saying, but in how we soften, divert, edit, or overcompensate. Sometimes this happens out of care. Sometimes out of habit. And sometimes, because the conversation we’re having isn’t quite the one we most need.

This experiment is designed to help you notice those quieter patterns. Not to fix them, and certainly not to analyse yourself into oblivion, but simply to see them. To name what has often gone unnamed.

And, like last time, you’ll have an AI co-pilot by your side. Not to give you answers, but to offer a second glance at the words you’ve already written.

Part One: A Single Thread

Start with a conversation that’s recent, and still emotionally live. It might be a WhatsApp thread with a close friend or partner. Or a work exchange that left a trace of something unspoken. Choose a stretch of messages where you’re the one doing most of the writing.

You don’t need to overthink it. Copy and paste 15–20 of your messages into your AI tool of choice (we used ChatGPT4o), and offer it this prompt:

“Based on these messages, what emotional patterns or habits can you see in how I communicate with this person? What might I be trying to say, but not quite saying?”

You might get some gentle surprises. You might get an uncomfortable truth or two. Either way, the insight is in the tone—not just the content. Are you reassuring more than necessary? Over-apologising? Offering clarity, but rarely asking for it in return?

If the response resonates, you’ve already learned something.

If it doesn’t, sit with that too. Sometimes, our resistance is a clue in itself.

 

Part Two: The Whole Thread

If you’re ready to go deeper – and we do recommend this only when time, energy and emotional bandwidth allow—you can repeat the process with a longer section of your message history. This takes more effort to prepare, but the richness of the insight can be worth it.

Export the full thread (on WhatsApp, for example, this takes a few taps: Export Chat > Without Media). Then either upload the full document or feed it in chunks. 

You can tell the AI:

• “This is a conversation thread between me and someone I message often. Please read this as one side of a relationship. What patterns can you see in how I communicate? Where might I be over-functioning, avoiding, caretaking, or softening? What’s not being said?”

And, if you really want to get under the surface:

• “What emotional needs might be present in how I write, even if I’m not stating them directly?”
• “What might I be hoping for, or assuming, in this relationship?”
•  “How might the other person experience me in this exchange?”

Sometimes, the results are a little startling. Sometimes, they’re just quiet affirmations of things you suspected but hadn’t yet put into words. And occasionally, they offer the courage to begin a conversation you’ve been meaning to have for a while.

Whatever shows up, let it be information, not indictment. The goal isn’t to catch yourself out. It’s simply to notice what’s there.

And to remember that all communication is relationship work, even when it’s just a few blue ticks and a waving emoji.

What Emerged (and Why It Mattered)

When I tried the experiment for myself, I expected something interesting, perhaps even clever. I didn’t expect to feel so thoroughly recognised.

The AI didn’t judge or pathologise. It simply noticed. It picked up on a set of relational rhythms I’ve played out over and over again in my life—this time, written in the cadence of texts and voice notes and off-the-cuff messages sent while cooking, walking, spinning plates.

What it noticed most wasn’t what I said. It was what happened just after I said something real.

A sentence about sobriety, followed by a flurry of jokes and music links. A moment of emotional honesty, quickly reframed. A confession, gently erased by humour. It described the pattern as “reveal, then retreat”—and that phrase has stayed with me.

“You’re comfortable being playful, affectionate, insightful, even soulful. But when conversations go toward your deeper vulnerabilities, there’s often a turn — to humour, distraction, logistics, or music. That’s not hiding. It’s more like editing for safety.”

Reading that was like being gently unmasked by someone who wasn’t trying to unmask me. It didn’t feel invasive. It felt generous. As if the machine had spotted something I’d been aware of, but not quite ready to name.

And then it went deeper. It noticed how often I offer praise, reassurance, affirmation. How much I want the people around me to know they’re doing brilliantly, that I see them, that they matter. But underneath that, it hinted at something else: the possibility that I tend to show up with open arms for others, even when I’ve quietly run out of fuel for myself.

There was a tenderness in how it framed that too:

“You clearly value being that safe space — it’s part of your identity. But your empathy runs outward more than inward.”

It’s common to read something like that and feel a mix of recognition and resistance, and that’s exactly what happened. A moment of: 

yes, I know this about myself. 

Followed immediately by: 

is that really a problem, though? 

And then a third layer: 

maybe that’s the point.

Then I had a tea and shut the laptop.  It was too much introspection for a superhot Monday. Incredibly valuable nonetheless!

 

An Invitation: Try It Yourself

There’s something strangely comforting about discovering that our patterns are visible. Not because we’ve been caught out, but because we’ve been seen. Softly, quietly, and without agenda.

This is the heart of the experiment. Not exposure, not analysis for its own sake, but an invitation to notice. To listen again to the conversations we’re almost having. To bring curiosity to the ways we care, and to the places where that care costs more than we realise.

If you decide to try this, don’t rush. Choose a thread that matters. Ask the questions slowly. Be open to the answers, even if you don’t act on them straight away. And if something resonates – if you feel the quiet jolt of recognition – sit with it. That’s where the good work begins.

Equally, if you feel the answers aren’t right – ask yourself why? AI is only as good as the questions you ask about the data you give it. 

You don’t need to do anything dramatic. You don’t need to send a big message, or start a long conversation, or fix anything all at once. You just need to pay attention. To what’s said, and what’s softened. To the silences. To the tone. To yourself.

And if you want to go further – if what shows up stirs something more interesting, we’re here to explore that with you. This is the kind of work we do at Lead Happy. Sometimes with AI. Always with care.

Relationships tend to thrive on presence rather than perfection, and presence, when you really stop to feel it, can start with something as simple as this.

What I found in this experiment wasn’t a new truth. It was a space to reflect on an old one, in a new light.

It reminded me that there are parts of us that speak, even when we’re silent. That emotional patterns don’t just show up in big declarations, but in punctuation. In which messages get sent, and which ones get saved as drafts. In how often we reassure, and how rarely we ask for reassurance in return.

And it left me wondering—what else might we be trying to say, without quite saying it?

Cue long, long rabbit hole. Watch this space 😉 

In the meatime, you can:

• Arrange a chat with us here
• Or you can Chat right away with Navianna, our AI team member. 
• Not read the first article yet?  Catch that here
• Want More of this but in shorter format?  No problems, subscribe to Anna’s Leadership Lowdown here
• Feeling in a coachy kinda mood?  Take a look and book a free session