Explainable AI–Assisted Therapy · a Wayward Robots framework

Therapy tools that explain themselves.

XAI Therapy™ is a transparency standard for AI-assisted practice — where every insight shows its evidence, the clinician stays in charge, and the client keeps their story. It’s the heart of everything I build.

Transparent. You see how it works. Explainable. Every insight has evidence. Interpretable. The data means something.
Why this matters

AI is already in the therapy room.
Most of it can’t tell you why.

Standard AI takes data and returns an answer — but no one, not even the clinician in the session, can see the reasoning behind it. In a field built on trust, informed consent, and doing no harm, an insight you can’t explain isn’t just uncomfortable — it’s a liability you can’t defend to your client, your supervisor, or your board. And therapy has a second, older problem: a clinician sees one hour of a client’s week — under 1% of a life — and rebuilds the rest from memory. The richest data in mental health care vanishes at the office door.

The difference, in one picture

Two ways AI can enter the therapy room.

One asks your therapist to trust an answer no one can see inside. The other shows its work at every step.

✕ The black box

How most AI works today

Session & client data goes in
? ? ?
hidden reasoning — sealed, proprietary
An answer comes out
No one — not even your clinician — can see why it said what it said.
The “reasons” it gives are a plausible story, not its real logic.
You can’t explain it to a client, a supervisor, or a board.
Result: trust it blindly — a liability
✓ The glass box · XAI Therapy™

How XAI Therapy works

1Client uses a tool built on a real clinical framework
2It produces structured, readable data — by design
3AI organizes it and cites the exact evidence — in the open
4 · You, the clinician, review and decide
An insight you can explain — every word of it
Every insight shows the evidence it’s built on.
The explanation is the data’s native shape — not bolted on after.
A human decides. You can defend every word.
Result: insight you can trust — you can see it

We don’t explain a black box after the fact. We never build one.

The standard

Three promises, kept out loud.

🔎

Transparent

You can see how it works — including where AI helped build it. Nothing hidden behind “proprietary.”

🧩

Explainable

Every insight traces to the client’s own inputs and the clinical framework behind them. Evidence you can point to and stand behind.

📈

Interpretable

The data means something clinically. A therapist reads it in seconds and acts on it in session — not decodes it later.

What “explainable” actually looks like

One tool, made visible.

A single cognitive-behavioral thought record — one of the tools in the framework. Watch what it captures.

🧠 CBT Thought Record · client-entered, framework-tagged
Situation
“Critical remark on my work in front of the team.”
Automatic thought
“Everyone thinks I’m incompetent.” — belief 70%
Distortion
Catastrophizing / mind-reading
Evidence against
“The same lead praised two other pieces this month.”
Reframe
“One piece of feedback is information, not a verdict.”
Belief, after
70%30%

A clinician reads this in four seconds: the belief lost more than half its grip, through evidence-checking, and here is exactly how. No inference from a fuzzy transcript. No trusting a machine’s word. That’s the work, made visible.

The XAI Therapy Framework

From a story a client wants,
to an insight a clinician can use.

Six steps. Each one keeps a human in charge and the reasoning in the open.

1

The story invites.

A living world draws the client toward the tool they need — one at a time. Engagement is how the medicine gets taken.

2

The tool does real clinical work.

Twenty-five years of frameworks — CBT, EMDR, self-efficacy, grounding — each an instrument that captures meaningful, structured data.

3

The data is glass.

Ratings, distortions, reframes, patterns over time — legible because the clinical framework is built into the tool.

4

AI works in the open.

It organizes, prompts, and calculates — citing the exact evidence behind each pattern. It drafts. It never decides.

5

The clinician decides.

Every insight is reviewed by a human who can explain it, override it, and own it. Human-in-the-loop is the design.

6

The client keeps their story.

Their experience and data belong to them — theirs to hold, understand, and take with them.

For the clinician

You stay the author.

XAI Therapy is built on one non-negotiable: the clinician is never replaced, only amplified. The AI hands you a sourced, framework-grounded draft — your judgment is the record. You finally see the life between sessions, and you can explain every bit of it.

Human-in-the-loop by design — nothing reaches a client without you.
Insights you can defend — to a client, a supervisor, a board.
Between-session signal, made clinical and readable.
Your client’s engagement, finally visible — and owned by them.

Read the full essay: The Glass Box →

Our vision — not yet proven, and we’ll say so

Where this is going.

In the spirit of the thing, the honest line between what is and what we intend: these are goals, not results we’ve measured yet. When we measure them, you’ll see the numbers and the method.

Forward-looking statements above are aspirations, not present-day claims. We’ll always tell you which is which.

Early access

Want the version of this future you can explain?

XAI Therapy is in early development, in closed pilots with a small group of clinicians. If you want the work to stay human — and want to see the reasoning — say hello.

Request early access