There is a particular kind of Sunday evening dread that therapists know well. The week ahead is full of sessions, each client needs different materials, and the hours between now and Monday morning are not enough. So you end up searching Pinterest for "CBT worksheet kids," printing something that is almost right but not quite, or improvising with a whiteboard and hoping the therapeutic concept lands anyway.

The preparation burden is real. Therapists who work with children and adolescents are not just clinicians — they are material designers, illustrators, game makers, and print shop operators. Every emotion card set, every board game, every social story, every visual schedule represents hours of work that happens outside of billable clinical time.

AI-powered tools are changing this equation. Not by replacing therapeutic expertise — that is irreplaceable — but by handling the production work that consumes therapist evenings and weekends.

The Time Problem

A 2019 survey of school-based mental health professionals found that clinicians spent an average of 5-8 hours per week on material preparation, documentation, and administrative tasks. For therapists in private practice who create custom materials, the number can be higher.

This time comes from somewhere. Often, it comes from the therapist's personal life, leading to burnout. Sometimes, it comes from clinical quality — the therapist uses a generic worksheet instead of creating the custom one they know would be more effective, because there simply is not time.

The promise of AI in this context is not about replacing the therapist's clinical judgment. It is about compressing the production timeline so that the custom material the therapist envisions can actually exist by Monday morning.

What AI Can and Cannot Do

Understanding the capabilities and limitations of AI tools prevents both overreliance and underutilization.

What AI Does Well

Visual consistency. AI image generation can produce illustrations in a consistent style across dozens of materials. The same character can appear on emotion cards, worksheets, and board games, creating visual coherence that would take a human illustrator significant time to achieve.

Rapid iteration. You can describe what you want, see a result in seconds, and adjust. "The character should look more worried, less scared." "Make the background warmer." "Show two children cooperating, not competing." This feedback loop that would take days with a freelance illustrator happens in minutes.

Layout and formatting. AI tools can generate print-ready layouts — proper margins, bleed areas, consistent typography — that would require graphic design skills to produce manually.

Content generation. AI can draft prompt cards for therapy games, generate scenario descriptions, create word lists, and suggest graduated exposure hierarchies. These drafts require clinical review and editing, but starting from a draft is faster than starting from blank.

Scale. Creating a 24-card emotion deck, a 60-question game card set, and a 5-page worksheet packet for three different clients no longer requires three different evenings of work.

What AI Cannot Do

Clinical judgment. AI does not know that this particular child needs emotion cards that include "numb" and "disconnected" because they are processing trauma. It does not know that the worry ladder should stop at "sleeping over at a friend's house" because that is the functional goal. Clinical decision-making is the therapist's domain, and AI should never make choices about therapeutic content without clinician direction.

Nuance in representation. While AI image generation has improved significantly, it can still produce images that are culturally insensitive, anatomically awkward, or emotionally ambiguous in ways that are clinically unhelpful. Every generated image needs a therapist's eye before it reaches a client.

Therapeutic relationship. Materials are tools in service of the therapeutic relationship. The most beautiful emotion card set in the world is useless without a skilled clinician who knows when to introduce it, how to facilitate the conversation around it, and when to put it away.

Evidence-based protocol fidelity. If you are implementing a manualized treatment (e.g., Coping Cat for anxiety, TF-CBT for trauma), AI-generated materials should supplement, not replace, the published protocol materials. The fidelity of the treatment depends on using validated materials as designed.

A Practical Workflow

Here is how the integration of AI tools into material preparation can work in practice.

Step 1: Clinical Planning (Therapist-Led)

Before touching any tool, define what you need clinically:

What is the therapeutic target?
What type of material best serves this target?
What specific content should the material include?
What developmental level and cultural considerations apply?

This step is entirely the therapist's domain. AI does not participate here.

Step 2: Content Drafting (AI-Assisted)

Use AI to generate first drafts of text content — worksheet prompts, game questions, scenario descriptions, psychoeducation text. Review every item for clinical accuracy, developmental appropriateness, and alignment with your therapeutic approach.

Expect to edit 30-50% of AI-generated clinical content. Some items will be excellent. Others will be vague, culturally tone-deaf, or clinically superficial. Your editing is what transforms a draft into a clinical tool.

Step 3: Visual Production (AI-Driven)

This is where AI saves the most time. Generate illustrations, design layouts, and produce print-ready files. Key considerations:

Describe your visual style clearly and consistently. If you want warm, hand-drawn illustration with soft colors, say that every time.
Use character persistence. Tools like Resource Builder allow you to define a character once and use it across all materials. This consistency matters clinically — children build familiarity with recurring characters.
Review every image. Look for unintended content, ambiguous expressions, and cultural misrepresentation. If something feels off, regenerate or edit.

Step 4: Clinical Review (Therapist-Led)

Before any material reaches a client, review the complete package:

Does every element serve the therapeutic goal?
Is the content developmentally appropriate?
Are the illustrations clear and unambiguous?
Would this material feel respectful and engaging to the client?
Does it meet your professional standards?

This review step is non-negotiable. Speed without quality is not a gain — it is a risk.

Step 5: Print and Prepare

Export print-ready files, print on appropriate cardstock, laminate if needed, and organize into your session materials. The physical quality of the material matters. A well-produced custom resource communicates care and professionalism to the child and their family.

Ethical Considerations

AI in clinical practice raises legitimate ethical questions that deserve thoughtful engagement, not dismissal.

Transparency. Should you tell clients (or their parents) that AI was used to generate the illustrations in their therapy materials? There is no professional consensus yet, but the principle of transparency suggests that if a parent asks, you should answer honestly. In practice, most parents care about the clinical quality of the materials, not the production method.

Bias. AI image generation models carry the biases of their training data. Be alert to stereotypical representations, particularly around race, gender, disability, and body type. If a tool consistently produces biased imagery, that is a reason to choose a different tool or provide more specific direction.

Data privacy. If you use AI tools that process text prompts through cloud services, ensure that no client-identifying information is included in your prompts. "Generate a worry ladder for an 8-year-old with separation anxiety" is fine. "Generate a worry ladder for [child's name] who is anxious about their parents' divorce" is not. This principle applies regardless of the tool's privacy policy.

Scope of practice. AI-generated materials should support evidence-based practice, not replace it. If you find yourself using AI to generate treatment protocols or clinical formulations, you have crossed a line. AI is a production tool, not a clinical supervisor.

When Not to Use AI-Generated Materials

There are situations where hand-made, personally created, or carefully curated published materials are more appropriate:

Trauma processing materials where every visual element needs to be carefully controlled for potential triggers
Materials for clients who are highly sensitive to "authenticity" — some adolescents will reject materials they perceive as "AI-made"
Published, validated assessment tools that must be used as designed for psychometric validity
Culturally specific materials where the nuances of representation require human cultural competence beyond what current AI can provide

Try It With One Session

Pick one upcoming session where you need a custom resource — an emotion card set, a coping worksheet, a feelings check-in tool. Use an AI tool like Resource Builder to produce it and compare the time investment to your usual process. Most therapists who try this don't go back to the old way.

The goal isn't to automate therapy. It's to automate the parts of therapy preparation that don't require a clinical degree, so you can spend your expertise where it actually matters: in the room with the child.

Creating Therapy Materials with AI: A Guide for Clinicians