Instructions

Working with AI

mentalhealthGPT doesn't mentalhealthGPT on generic responses; instead, it mentalhealthGPT with the actual context of your work—sessions, documents, notes, and your professional methodology.

Five things you should know

How to Work Effectively with AI

🎯
Choose the right space for thinking

mentalhealthGPT two different workspaces: TheChat you engage in free professional thinking—independent of any specific case file. TheChat with the full context of a case: sessions, documents, notes, and your professional methodology.

WorkspaceWhat it is particularly useful for
ReflectionChat For free professional reflection, hypothesis-building, shifting perspectives, critical analysis, and bias checks—deliberately without the context of a dossier.
DossierChat For case-specific questions based on specific meetings, documents, notes, and active context libraries.
Chat: “What alternative explanations are there for withdrawal in therapeutic relationships?”
DossierChat: “What forms of withdrawal are evident throughout this specific dossier?”

💡 Make deliberate use of both spaces: TheChat you work closely with the specific material. TheChat you gain some perspective, test hypotheses, and develop alternative viewpoints.

🔍
Use session analysis

InChat , youChat the "Analyze" menu at the top—select one of the five analysis styles to immediately see the results in the modal. The results are not saved and can be copied directly.

When is it particularly worthwhile?
After complex or ambiguous sessions In the event of a strong countertransference reaction Before Supervision Before writing a report When something seems important but remains unclear

The Five Analysis Styles — Which One Is Right for What?

  • Brief notes — 3–5 sentences on the main topic and overall tone. Ideal for quick documentation right after the meeting.
  • Summary — Organized into four sections: Concerns / Topics / Interventions / Next Steps. The standard clinical format for progress documentation.
  • Supervision Question — A single open-ended reflective question for follow-up. Particularly helpful for preparing for Supervision.
  • Hypotheses — 2–3 clinical hypotheses regarding dynamics, patterns, or relationships — expressed in observational terms, not as a diagnosis.
  • Chat Topics — A bulleted list of recognizable conversation topics, with a maximum of seven items. For a quick overview and to take notes on the conversation.

💡 Most effective immediately after the meeting—while nonverbal cues are still fresh in your mind and the transcript is still new.

Five Principles

How to Get Better Answers

Good clinical reflection doesn't come from better AI—it comes from better questions. These five principles can help, even without technical background knowledge.

🎯

Specific rather than vague

The AI can only respond to what you actually ask. The more specific the question, the more clinically useful the answer.

"What do you think of this meeting?"
“What relationship patterns have been recurring over the last three sessions?”
📐

Specify the desired format

When you define the desired output, you’ll get results that are ready to use—no post-processing required.

"Summarize the key stress factors in 150–170 words."
"Write a factual progress report in a professional tone."
"List possible differential hypotheses in bullet points."
🔭

Set the perspective

A specific analytical lens yields more focused answers—and prevents the AI from giving generic responses.

“Analyze the meeting from the perspective of relationship dynamics.”
“What signs point more toward anxiety regulation than toward opposition?”
“What aspects would be Supervision relevant for Supervision ?”
📈

Use History and Compare

That is the real value of long-term case files: they reveal patterns over time that remain hidden in individual cases. Questions spanning multiple sessions offer a different perspective than those limited to a single session.

“What issues have changed since the first meetings?”
"What patterns recur when under stress?"
"What specific progress can be demonstrated in the material?"
🗂️

Explicitly require the structure

Explicit structural guidelines make outputs directly reusable—for reports, Supervision notes.

"Answer using three hypotheses, each with a corresponding counter-hypothesis."
“Separate observation, interpretation, and open questions.”
"Summarize the meeting in no more than five key points."

📚
Incorporate your own methodology

AI doesn't replace your methodology—it builds on it.

When you save therapeutic frameworks, protocols, or institutional guidelines in the context library, AI responses will be more closely aligned with your actual workflow—rather than generic recommendations.

What you can include: therapeutic models (e.g., CBT protocols, schema therapy, attachment-based approaches), institutional standards, your own work guidelines, or forms that you use regularly.

Usage model

Understanding the Role of AI

mentalhealthGPT a tool for reflection and structuring—not for making decisions on its own.

AI helps with

  • Reflection and Organization
  • Pattern recognition across multiple sessions
  • Formulating and refining hypotheses
  • Formulation and Processing
  • Shifting perspectives and identifying alternatives

Your responsibility remains

  • Technical expertise and sound judgment
  • Final clinical decisions
  • Building Relationships with Clients
  • Ethical and Legal Responsibility
  • Assessment of Risks and Urgency

mentalhealthGPT reflection, structuring, and clinical reasoning. The professional assessment and final decision remain your responsibility.

Practical tip

When AI is particularly helpful

Many users see the greatest value in these situations:

Complex cases Extensive case files involving numerous meetings Preparing reports and certificates Supervision Understanding recurring dynamics Developing differential hypotheses Patterns across multiple sessions Structuring diffuse situations Structuring ambivalence Formulation assistance for reports Reflecting on dead ends Reviewing therapy goals