IFS Between Sessions: Staying With Parts Outside the Room

Internal Family Systems is the model I see clinicians reaching for hardest right now, and for good reason — it gives clients a way to relate to their own experience that's humane, precise, and oddly portable. But it has a between-session problem that the trainings don't quite name, and a between-session gift that almost nothing else in our field offers. Both live in the same place: the 167 hours a week the client spends outside your office.

This is a clinician-to-clinician look at what actually travels out of the room — and what to give a client so a part doesn't run the whole week unwitnessed.

The between-session problem IFS doesn't name

In session, the work is beautiful and it works. You help the client notice a protector, ask it to give a little space, find some Self-energy, turn toward the part with curiosity. There's real movement. They leave regulated, unblended, in relationship with their system.

Then it's Tuesday at 11pm, the old trigger lands, and a part takes over completely. Here's the hard structural fact: in that moment there is no Self present to do the unblending, and no therapist in the room to help them find it. The client doesn't have a part right then — they are the part. The very capacity IFS depends on, Self leading, is the capacity that goes offline exactly when the part fires.

This is the same shape as why insight doesn't stick in general: the work was encoded in a regulated, Self-led state, and it has to be retrieved from a blended, activated one. Asking a blended client to "go inside and find Self" is asking the part that's driving to hand over the keys. Sometimes it works. Usually it doesn't, and the client concludes they're bad at IFS.

...and the gift IFS has that almost nothing else does

Now the good news, because it's genuinely good. Parts language is portable in a way most clinical vocabulary is not. A client who can say to themselves, alone, "okay — the manager is online right now, and there's a young part underneath it that's scared" is doing real, first-person clinical work without you in the room. They have a handle on their inner experience that doesn't require them to remember a worksheet or perform a technique.

That's the asset to build on. Most models give the client a concept to recall. IFS gives them a relationship to notice. The between-session task, then, isn't to get the client doing full sessions alone. It's to make the noticing reachable in the activated state — and to keep it pointed back at you.

What actually helps a client stay with their parts outside the room

A few principles hold up across caseloads. They're deliberately humble, because humility is what survives the 167 hours.

1. Aim for noticing the blend, not achieving Self. The realistic between-session win is "oh — a part is online," caught a little earlier each week. Full Self-leadership at 11pm is a high bar most clients can't clear alone, and setting it as the goal just manufactures another way to fail.
2. Use the client's own name for the part, built in session. Their word, their image, the part's age and role as they describe it. "The foreman," "the little one by the window," "the one that scrolls." A client can reach their own name for a part while flooded; they cannot reach "your protector part" in your clinical phrasing.
3. One degree of unblending, not the whole protocol. A single slow breath, attention turned toward the part, "I see you, I'm here." That sliver of distance — enough to say "a part is driving" instead of being fully taken over — is plenty. It's the difference between zero space and some space, which is the only difference that matters in the moment.
4. Protect the relationship — route it back. The between-session practice is not a substitute for the unblending you do together; it's what the client carries in. If a practice pulls the client toward doing IFS alone instead of bringing the material back to you, it's quietly working against the therapy.

A simple between-session parts practice

Here's the shape I use. It's four steps, it's built with the client in session, and it assumes nothing about their state except that a part will show up when you're not there.

1. Name the part together before they leave. Land on the client's own name for it and write that exact phrase down — in their language, not yours.
2. Decide the tell. Identify the body signal or situation that usually means this part is about to take over: the tight chest, the 11pm scroll, the urge to send the text. That's the cue to notice — not a task to complete.
3. Notice and name in the moment. When the tell shows up, the client names it — "the foreman is online" — and takes one breath turned toward the part. The noticing is the whole win.
4. Carry one line back. Jot a single line — what fired the part, what it seemed to be protecting — to bring to next session. That line is often worth more than the homework ever was.

Notice what this is not: it's not journaling, not a meditation series, not a technique to master. It's a way to keep one thread of Self-awareness alive in the hours the part would otherwise have entirely to itself.

What tends not to work

Worth naming plainly, because these are the defaults and they underdeliver.

• Long Self-led meditations. Lovely in theory, rarely opened mid-flood. They assume the regulated state they're meant to rescue the client from.
• Generic "parts journaling" worksheets. In the clinician's language and structure, which is the last thing reachable when a part is blended.
• Direct-to-client parts apps that go around you. Some are thoughtful, but the ones built to do parts work with the client unsupervised cut the clinician out of the loop. The U-turn that matters clinically is the one that comes back to the room, not the one that keeps the client in the app.

The line — a mirror, not a parts-chatbot

There's an obvious temptation to hand the client something that talks back at 11pm, and it's worth being careful. A chatbot that role-plays the therapist or claims to "do IFS with" the client unsupervised is the wrong category — general-purpose AI chatbots respond appropriately to people in distress below 60% of the time, against around 93% for licensed clinicians, and that gap is structural, not a tuning issue.

What's defensible is narrower: a mirror that reflects "a part seems to be online for you right now," in the client's own language, and routes that noticing back to you. It reflects; it doesn't interpret, treat, or hold clinical responsibility, and it keeps you in the loop. Anything touching a client's material has to be HIPAA-compliant by design, and anything that goes around the clinician straight to the client has crossed into a different and riskier lane. I draw that line in more detail in AI as clinical tool, not replacement, and the broader case for between-session work is in the pillar on keeping clients engaged between sessions.

FAQ

Can clients do IFS on their own between sessions?

Partly, and it matters which part. Noticing — catching that a part is online and naming it in their own language — is genuinely doable alone and is most of the between-session value. Deep unblending and getting to know an exile usually need you, because when a part takes over there's no Self present to do the work. So aim the between-session practice at noticing and one degree of space, not at full Self-led parts work without you.

What's the difference between a part being "online" and the client being blended?

A part being online means a part is active and influencing how the client feels and acts. Blending means the client has merged with it — they aren't experiencing the part, they are the part, with no distance from it. The between-session goal is to help the client catch the blend a little earlier each week and create even a sliver of space, so they can say "a part is driving" instead of being fully taken over.

How do I help a client stay with a part when they're flooded?

Don't ask for the calm, Self-led state that's offline when they're flooded — that's a state-access problem, not a motivation one. Give them something tiny that works inside activation: their own name for the part, a single breath turned toward it, one degree of unblending. Build it in session so it's already theirs, and keep the bar at noticing rather than resolving.

Is it safe to use an AI tool for parts work between sessions?

It depends on the category. A chatbot that role-plays the therapist or "does IFS with" the client unsupervised is the wrong tool — general-purpose AI chatbots respond appropriately to distress below 60% of the time versus around 93% for licensed clinicians. A HIPAA-compliant mirror that reflects "a part seems to be online for you, in your words" and routes it back to you is a different thing. Safety lives in the design: HIPAA-compliant, clinician-in-the-loop, not a treatment substitute.