Between Therapy Sessions: The Unexpected Role AI Is Starting to Play

Technology is changing quickly.

And like most major changes, AI comes with benefits, concerns, questions, risks, and a whole lot of opinions.

There are conversations to be had about privacy, accuracy, ethics, overreliance, and even the ways AI is changing how we work. As a therapist, I’ve even seen benefits when it comes to things like documentation and administrative support—with appropriate attention to HIPAA, privacy, and removing identifying client information.

But that’s not what this particular blog is about.

I’m also not here to convince anyone that AI is good or bad, that everyone should use it, or that it belongs in everyone’s mental health journey.

I’m simply sharing something I’ve started noticing in my work as a therapist—and something I’ve experienced in my own life:

For some people, AI is becoming a useful tool in the space between therapy sessions.

And I think that’s worth talking about.

Life Doesn’t Wait for Your Next Therapy Appointment

Most people in traditional outpatient therapy see their therapist once a week. Some every other week.

That leaves a whole lot of life happening outside of that 50-minute session.

You leave therapy Tuesday and get into an argument with your mom Wednesday.

Something happens at work Thursday.

Someone disappoints you Friday.

You get a text Saturday that completely triggers you.

And your next therapy appointment isn’t until Tuesday.

Your therapist isn’t available every time something happens.

Nor should they necessarily be.

Healthy therapeutic relationships have boundaries. Unless someone is receiving a more intensive level of care, therapy generally isn’t a 24/7, on-call relationship.

Ideally, we also have healthy support systems outside of therapy.

But here’s the reality:

Not everybody does.

Some people don’t have someone they feel emotionally safe enough to be vulnerable with.

Some people have support, but that person isn’t available.

Sometimes the people around us aren’t equipped to help us process what we’re experiencing.

And sometimes we’ve talked our friend’s ear off about the same situation for the tenth time and our support system needs a break too.

That’s human.

So what happens in the meantime?

That’s one of the places where I’ve started seeing AI used in an interesting way.

Not as the therapist.

Not as the friend.

Not as the solution.

As a tool.

Sometimes You Just Need a Buffer

One of the biggest potential benefits I’ve noticed is something very simple:

The pause between feeling something and doing something.

Something happens.

You get angry.

You’re hurt.

You’re triggered.

You want to send the text.

You want to make the phone call.

You want to tell somebody exactly how you feel right now.

And sometimes what we need first is somewhere for all of that emotion to go.

I’ve watched clients use AI almost as a buffer during those moments.

Instead of immediately reacting to the person or situation, they put the thoughts somewhere else first.

“This happened. I’m pissed. This is what they said. This is what I want to say.”

Sometimes just explaining the situation forces you to slow down enough to begin organizing what actually happened.

Then maybe the question becomes:

“Okay, how can I communicate this without being aggressive?”

Now there’s a pause.

Something happened → I felt something → I paused → I decided how I wanted to respond.

That pause matters.

Because emotional regulation doesn’t mean we stop having emotions.

It means our emotions don’t automatically have to determine our next behavior.

And I’ve Used That Buffer Too

I also don’t want to write about this as though I’m standing on the therapist side of the room observing what everybody else does.

I’m a therapist.

I’m also a human being.

And I’m in therapy myself.

My emotional moments don’t conveniently schedule themselves for right before my therapy appointment either.

I’ve had situations happen where I’ve been angry, hurt, overwhelmed, disappointed, or simply trying to understand why something affected me as strongly as it did—and my therapist wasn’t available in that exact moment.

Nor would I expect them to be.

I’ve used AI during some of those moments too.

Sometimes I just need somewhere to put everything I’m thinking before I respond to another person.

Sometimes I’ll essentially say:

“Okay. This happened. I’m really upset. Why might this be bothering me this much?”

And sometimes seeing different possibilities reflected back helps me slow down enough to realize:

Oh.

I’m not just angry.

I’m actually hurt.

Or disappointed.

Or I expected something different from that person.

Or something about the situation touched a deeper emotional wound.

Does that mean AI gets to tell me what I’m feeling?

No.

Does that mean everything it suggests is correct?

Absolutely not.

I still have to think.

I still have to ask myself, Does that actually fit my experience?

And sometimes what I discover through that process becomes something I bring to my own therapist later.

That’s an important distinction for me:

I’m not handing a tool responsibility for my emotional experience. I’m using a tool to help me slow down and examine my emotional experience.

And that’s very similar to what I’ve started seeing some clients do.

When You’re Dealing With Really Difficult Emotions

I’ve worked with clients navigating incredibly stressful systems. DCFS is one example.

Imagine something happens with your case.

You’re frustrated.

You’re scared.

You’re angry.

You feel unheard.

You feel like decisions are being made about your family and nobody understands what you’re experiencing.

Those emotions make sense.

But immediately calling your social worker while you’re highly dysregulated and screaming at them probably isn’t going to help you accomplish what you’re trying to accomplish.

That’s where the buffer can become useful.

Instead of immediately making the call or sending the email:

“This just happened and I’m extremely angry. This is what I want to say.”

Get it out.

Then maybe:

“What am I actually trying to communicate here?”

Or:

“How can I communicate my concerns assertively without attacking this person?”

The situation hasn’t magically disappeared.

The emotion hasn’t been invalidated.

But now there’s some space between the emotion and the action.

And when that client comes back to therapy?

We can still process it.

AI didn’t replace the therapeutic work.

It helped create a bridge until we could continue the therapeutic work.

“I’m Angry” Usually Isn’t the Whole Story

Another place I’ve seen AI become useful is emotional identification.

A lot of people can recognize:

I’m mad.

Okay.

But what’s underneath the anger?

Were you hurt?

Did you feel rejected?

Were you embarrassed?

Disappointed?

Dismissed?

Unimportant?

Scared?

Did something happening today remind your nervous system of something much older?

These are conversations we absolutely have in therapy.

But emotional awareness is a skill.

And people don’t magically master a skill because we talked about it for 50 minutes on Tuesday.

They have to practice it while they’re actually living their lives.

So I’ve had clients come back and say something happened during the week and they put the situation into ChatGPT because they couldn’t figure out why they were so upset.

Maybe through that process they realized:

“I think I actually felt rejected.”

Great.

Now let’s explore that.

Why did rejection come up?

What did you make the interaction mean about you?

Is there a pattern here?

What did you need in that moment?

That’s where therapy takes it deeper.

The AI didn’t complete the therapeutic work.

It gave us something to bring into the therapeutic work.

Practicing Communication Before the Real Conversation

This might be one of the areas where I’ve seen the most practical benefit.

Communication is something I work on constantly with clients.

Some people have spent their entire lives people-pleasing.

Some avoid conflict.

Some become passive.

Some stay quiet, quiet, quiet—and then eventually explode.

Others know exactly what they want to communicate but struggle to express it without becoming aggressive.

We’re trying to find that middle ground:

Assertiveness.

Imagine something happens Wednesday and you need to talk to your mom.

You’re working in therapy on not people-pleasing anymore.

You know you need to say something, but you don’t know how.

Instead of avoiding the conversation completely—or going into it emotionally activated—you might ask:

“I need to talk to my mom about what happened. I’m working on not being passive anymore, but I also don’t want to be aggressive. How could I approach this assertively?”

Now you have a starting point.

And I’ve had clients bring those conversations back into therapy.

We’ll look at what was suggested.

We’ll change things.

I’ll ask:

“Does that actually sound like you?”

Then we’ll role-play.

“Okay. I’m your mom. Say it to me.”

Maybe they stumble through it.

We try again.

Maybe we realize the wording doesn’t actually communicate what they need.

We adjust it.

And eventually they have the real conversation themselves.

That’s the important part.

The tool didn’t have the conversation for them.

It helped them practice a skill they are learning in therapy.

Validation Can Help—But We Also Have to Talk About Dependence

This is where I think the conversation needs nuance.

Sometimes people need validation.

Especially people who haven’t had emotionally safe relationships.

If you’ve spent years being told you’re overreacting, you’re too sensitive, your feelings don’t matter, or you’re always the problem, having something reflect back:

“It makes sense that this affected you.”

can feel incredibly relieving.

But I also think we have to be careful.

Because if every uncomfortable emotional experience turns into:

“ChatGPT, tell me I’m right.”

“Tell me they’re wrong.”

“Are my feelings valid?”

“What should I do?”

then we’re entering different territory.

Now the tool may be becoming another source of external validation that someone feels unable to function without.

And I don’t want my clients dependent on AI for validation.

But here’s something equally important:

I don’t want my clients completely dependent on me for validation either.

I don’t want them dependent on their partner, best friend, parent, or anyone else to regulate every emotional experience for them.

And personally, I don’t want to become dependent on an outside source for that either.

Because the goal is eventually developing the ability to say:

I know what I’m feeling.

My feelings make sense.

I can tolerate this emotion.

I don’t have to respond immediately.

I can calm myself down.

I can figure out what I need.

I can decide how I want to communicate it.

That’s the work.

AI Isn’t Your Therapist—and It Isn’t Your Friend

I think we also have to be very clear about what AI isn’t.

AI isn’t a therapist.

It isn’t your friend.

It doesn’t replace genuine human connection.

Relationships require things technology can’t replicate.

Vulnerability.

Reciprocity.

Boundaries.

Repair.

Learning how to tolerate disagreement.

Being disappointed by someone and figuring out how to navigate it.

Allowing another human being to actually know you.

If someone begins using AI instead of developing relationships because human connection feels uncomfortable or unpredictable, that’s something worth paying attention to.

There’s a big difference between a tool supporting human connection and a tool replacing human connection.

The distinction matters.

The Goal Was Never the Tool

So no, this isn’t me telling everyone to start processing their emotions with ChatGPT.

I’m simply sharing something I’ve started noticing.

I’ve watched clients use AI in ways that genuinely supported skills we were already working on in therapy.

I’ve also experienced moments in my own life where it helped me slow down, reflect, and organize what I was feeling until I could process it further on my own, with someone I trust, or in therapy.

Sometimes we don’t need something to solve the problem for us.

Sometimes we need something that helps us pause long enough not to make the problem bigger.

And I think that’s the part of this conversation that gets lost when we reduce AI to being either “good” or “bad.”

It’s a tool.

And tools should support our growth—not become responsible for it.

If it helps you pause instead of react, that’s useful.

If it helps you recognize that underneath your anger you’re actually hurt, that’s useful.

If it helps you organize your thoughts before a difficult conversation, that’s useful.

If it gives you something meaningful to bring back into therapy, that’s useful.

But ultimately, I want the same thing for my clients that I want for myself:

To recognize our own emotions.

To regulate ourselves.

To trust ourselves.

To communicate what we need.

To validate ourselves.

And to know when we genuinely need support.

Because whether the tool is therapy, journaling, medication, a coping skill, an app, AI, or another form of support:

The goal was never to become dependent on the tool.

The goal is for the tool to help us build skills within ourselves.

And sometimes, a little support during those six days between therapy sessions can help us practice becoming the person we’re working toward becoming.