It's Not AI Anxiety. It's Grief You Can't Explain Yet.

A friend texted me last week. The message was short:

"They replaced my function with a custom GPT."

He's 31. Senior engineer at a Series B. We spent three days talking after that message. What he kept circling back to wasn't anger. It wasn't fear. It was a sentence I haven't been able to stop thinking about:

"I don't know how to feel about something that hasn't fully happened yet."

That sentence is what AI job anxiety actually feels like — and it's not anxiety at all. Anxiety is about something that might happen, sometime, maybe. What he was describing is something older and stranger. A slow goodbye to a version of his career that already started disappearing.

The word for that exists. Most people just don't use it for jobs.

I want to talk about why this kind of stress hits different. What the body is doing under it. And why the standard career advice — "just upskill," "learn to use AI," "adapt or get left behind" — never quite reaches the part of you that actually needs help.

It's Not AI Anxiety. It's Anticipatory Grief.

Here's the reframe that changes everything: AI job anxiety isn't anxiety in the traditional sense — it's anticipatory grief. It's the body grieving a version of your career that's been quietly disappearing in real time. Unlike typical workplace stress, it doesn't resolve with a job change. The threat is category-level: the role itself is shifting, and the nervous system has noticed.

Anticipatory grief is a real psychological concept. It was originally named in research on terminal illness — the way someone begins to mourn a person they haven't lost yet. The body starts processing the absence before the absence is real.

The structure fits AI displacement almost exactly.

You're not losing a job. You're losing a version of work. A version of your identity. A version of the next twenty years of your career that you had quietly assumed would still exist. The thing you're mourning hasn't fully happened — and may never happen the way you fear — but the direction of the change is undeniable, and the body has filed it.

This is why AI anxiety feels so disorienting. You're trying to use anxiety language for an experience that's actually closer to grief. The two states feel different in your body, last different lengths of time, and respond to completely different interventions. Treating one like the other is part of what makes it feel so hard to name.

The first thing that helps is the word.

The Body's Response to a Slow Loss.

When you grieve something, your body knows.

In acute grief — when you lose someone, when you lose a job, when something specific ends — the physiological pattern is dramatic and short-lived. Cortisol spikes. Sleep breaks. HRV plummets. Within weeks, the system begins to find its way back to baseline. The loss is real, the body responds, and the body recovers.

Anticipatory grief is different. The loss hasn't fully arrived, so the body never gets the signal that it's over. It just stays in a low-grade alert state, week after week, sometimes month after month — keeping watch for an event that keeps not quite happening.

This is what we're seeing in early data on people who report sustained AI-related career stress:

• Resting heart rate trends two to five beats higher than baseline, for weeks

• HRV stays flat — not a sharp drop, but a quiet ceiling that doesn't lift

• Sleep latency creeps up by ten to twenty minutes — falling asleep takes longer, but you don't always notice the trend

• A specific pattern shows up around evening Slack: a small spike when a new headline about AI hits, then a slow, incomplete return to baseline

None of these signals are dramatic. That's what makes them dangerous. They're below the threshold most people would notice, and they compound across months in a way no single day can explain.

You don't feel sick. You feel a little worn down. A little less sharp. A little less able to focus on the long-horizon things that used to come naturally. You assume you just need a vacation.

You don't need a vacation. You need a body that hasn't been processing a slow career loss for the last eight months on a loop.

The pattern is what makes this category of stress particularly hard to catch. Most stress conditions have an obvious trigger and an obvious resolution — you can point at the thing that's wrong, address it, and watch the body recover. Anticipatory grief about a job category has neither. There's no event to point at. There's no resolution to chase. There's only the slow accumulation, day after day, of micro-spikes that never fully clear.

By month six, most people have rewritten their internal baseline. The slightly elevated heart rate feels normal. The slightly worse sleep feels normal. The slightly diminished sharpness feels normal. They don't realize anything has shifted because the shift was too gradual to notice from inside it. This is the cost nobody is talking about.

Why AI Anxiety Feels Different Than Layoff Stress.

Layoffs are bounded. AI replacement is not. This is the difference nobody is naming, and it's the difference that matters most physiologically.

A layoff has a shape. It's specific. It happens to a specific person, on a specific day, with specific consequences. There's a moment when it's over — when you've signed the paperwork, talked to your friends, filed for unemployment, started the next thing. The system gets a clear signal: the event has occurred, you can begin recovery.

AI replacement doesn't have a shape. It's not happening to one person — it's happening to a category. It's not on a specific date — it's spread across years of incremental capability gains. It doesn't end — every model release, every new agent, every Twitter thread reopens the question.

You can't recover from something that hasn't fully happened. The body literally doesn't know what to recover to.

This is also why the usual career-stress framings don't fit:

• "Get a new job" — assumes there are jobs in your category that aren't moving in the same direction

• "Pivot industries" — assumes there's a stable industry to pivot into

• "Wait for the market to come back" — assumes this is a cycle, not a structural change

None of those provide the closure the nervous system is looking for. They're all rational moves that may or may not solve the economic problem — but they don't solve the physiological one. The body is still grieving the original career version. The new role just inherits the same low-grade alert state.

What operators and senior knowledge workers are starting to discover, the hard way: the AI question doesn't resolve by switching companies. It resolves by your body finally being given a signal that the immediate threat has cleared. And that signal has to come from somewhere — usually not from your career.

Upskilling Doesn't Reach the Nervous System.

Every AI-anxiety thread on LinkedIn ends the same way. Upskill. Learn the tools. Become AI-native. Adapt or get left behind.

This advice isn't wrong. It's just incomplete in a way nobody talks about: it treats the problem as informational when it's actually somatic.

You can learn every model in the OpenAI lineup and your HRV will not change. You can become the best prompt engineer at your company and your sleep latency will still creep. You can pivot into AI-adjacent work and your body will still be processing the loss of the version of your career it spent fifteen years building.

The reason is simple: knowledge updates the thinking brain. The thinking brain was never the one in distress.

The system in distress is the autonomic one — the part that runs your heart, your breath, your sleep, your background sense of safety. It doesn't read Twitter threads. It doesn't take courses. It responds to one thing: signals about whether the immediate threat has cleared.

When those signals don't arrive, the system stays elevated. Forever, if needed. It doesn't have a quit button.

So you find yourself in this strange place: doing all the "right" things, learning all the tools, getting all the certifications — and still feeling exhausted in a way the productivity metrics can't explain. You assume you're not doing enough. You add more. The exhaustion deepens.

You're not failing at upskilling. You're succeeding at it. The problem is that upskilling never had the keys to the room where the actual stress lives.

I've watched this happen to people I respect. A designer who took six AI courses and learned every tool — and still couldn't shake the low-grade dread that her field was disappearing under her. An engineer who built half a dozen agents and shipped them to production — and still woke up at 4am wondering if the version of his career he had imagined was already gone. A copywriter who became the AI champion at her company — and still felt, every Sunday night, a small specific weight she couldn't explain to her partner.

None of them are doing it wrong. They're doing exactly what they were told to do. The "right" actions aren't reaching the part of them that needs reaching, and the gap between effort and relief is what's draining them.

What the body needs isn't more information. It's intervention — short, repeated, in the moment the spike is happening. A signal that the immediate threat has cleared, even if the long-term picture hasn't. A way for the system to come back to baseline so it can hold the next spike, and the next.

That's not something a course can do. It's something a wearable can.

What Naming It Changes.

Here's the part of the research most people miss: naming an emotion changes its physiological intensity.

When you label what you're feeling — even silently, even just to yourself — the amygdala (the brain's threat detector) measurably reduces its activation. The body shifts from "unknown threat present" to "threat identified," and the entire stress response begins to relax slightly. This is one of the most replicated findings in affective neuroscience, and it's quietly relevant to AI anxiety.

If what you're feeling is anxiety, you've been trying to treat the wrong thing.

If what you're feeling is anticipatory grief, the door opens.

It doesn't fix the macro situation. The AI question is still real. Your industry is still shifting. The slow loss is still slow. But the body can now do something it couldn't do before — file the experience as the thing it actually is, instead of a vague unease without a name.

And once it's filed, the next move becomes possible: interrupting the loop.

The reason micro-interventions work for this kind of stress is that they don't try to resolve the unresolvable. They just give the body a clean, brief signal — the immediate threat has cleared, you can stand down for a moment — and let the system find its way back to baseline. Sixty seconds, repeated across the day, every day. Compound effect over a year.

This is the part Momomoon was built for. Not the obvious crises. The slow, unending kind that nobody else has language for — the kind that pools in the nervous system while you're "just upskilling" and "just learning the tools" and quietly losing baseline.

If you've been off in a way that didn't fit the word anxiety, you might have been grieving something nobody told you it was okay to grieve.

Now you have the word.

That's the first thing that helps.

Momomoon is a stress intervention system, not a tracker. It runs on your wearable — detecting rising stress through HRV and context signals, then delivering a haptic nudge that guides a 1–2 minute recovery reset. No screen. No dashboard. Just an early signal and a brief reset, at the moment your nervous system needs it most.

Reserve Early Access for $1 →