PARTS Prompt Generator — an intelligent tool for creating pedagogical prompts
By the CPO Creativity team | 10 November 2025, 2:27 pm
1. Context and rationale
In 2025 Google published the LearnLM Partner Prompt Guide, presenting the integration of the LearnLM model in Gemini 2.5, the first large language model fine‑tuned on pedagogical principles. The document emphasises that effectively using artificial intelligence in education requires a clear pedagogical structure in the prompt, not just a technical instruction.
Google’s recommended framework is P·A·R·T·S, where:
- P – Persona: the role of the AI model (e.g., “You are a literature teacher…”);
- A – Act: the action it should perform (e.g., “Create a lesson” or “Explain the concept”);
- R – Recipient: the target group (students, teachers, etc.);
- T – Theme: the topic or concept;
- S – Structure: the format or framework for the output (e.g., the 5E model, UDL principles, CER exit ticket).
This framework gives the model clear context, role and didactic direction, leading to consistent and realistic responses comparable to a real teacher.
2. The idea behind the generator
The PARTS Prompt Generator — Pro was developed as a universal tool for creating pedagogical prompts based on the same principles described in the LearnLM document. It targets:
- Teachers and trainers who want to create clear, structured instructions for AI;
- EdTech developers implementing Gemini, OpenAI or Copilot in educational systems;
- University lecturers and methodologists testing different approaches to teaching with AI.
The generator combines instructional clarity with pedagogical behaviour — not just producing text but modelling interaction between teacher and intelligent assistant.
Configurator (P·A·R·T·S)
The tool presents configurable fields for Persona, Act, Recipient, Theme and Structure, compatible with Gemini, OpenAI and Microsoft Copilot. Users can choose roles (e.g., literature teacher, biology teacher), actions (create a 5E lesson, adaptive test), recipients, themes and structures (UDL + rubric, role‑playing scenarios). There is also an optional field for context/sources.
Pedagogical checklist
Six key LearnLM principles are built into the checklist:
- Active learning;
- Cognitive load management;
- Metacognition;
- Curiosity;
- Adaptability;
- Safety and correctness.
Each principle can be toggled on and is automatically included as a behavioural instruction in the generated prompt.
Quick templates (“chips”)
The generator provides pre‑built templates for roles, actions and structures — for example, literature teacher, biology teacher, create a 5E lesson, rewrite a text, UDL + rubric or role‑play scenario + reflection.
Quality metrics (KPI panel)
A real‑time KPI panel shows the length of the prompt, number of activated pedagogical principles and an overall quality score based on the level of structuring.
Interactive output
Users can switch between four output views: a universal PARTS prompt, formats tailored for Gemini, OpenAI and Copilot. This allows quick testing and transfer of the finished prompt to the chosen environment.
4. Pedagogical value
The LearnLM guide stresses that a quality prompt equals quality learning. A good prompt should encourage curiosity and self‑reflection, provide scaffolding without giving away the answer, manage cognitive complexity with clear structure and adapt tone and difficulty based on feedback. The PARTS Prompt Generator incorporates these principles not just in theory but as functionality. Each generated prompt can be used directly in Gemini AI Studio, Microsoft Copilot or ChatGPT without extra tuning.
5. Educational applications
The generator supports diverse educational uses:
- Teacher: creates an adaptive lesson or test based on the 5E model for a specific class.
- Methodologist: analyses the pedagogical effectiveness of the prompt and engagement level.
- EdTech developer: integrates PARTS prompts into LMS or interactive modules.
- Education student: learns how to create structured didactic instructions for an AI assistant.
- University: conducts experiments on models for metacognitive learning.
6. Significance for the future of AI in education
The generator illustrates how Bulgarian pedagogical practice can build on global standards set by Google LearnLM. It transforms AI from a mere answer generator into an intelligent pedagogical partner that thinks like a teacher, asks questions like a mentor and supports like a colleague.
7. Conclusion
Through the PARTS Prompt Generator, educators gain a new language for dialogue with AI — a language that is didactic, adaptive and humane. This marks a step toward the next era of learning through artificial intelligence, where prompts are not just commands but acts of methodology.
