The journal can also be used without AI: you can fill it in, review it and draw your own conclusions by hand. This page is for anyone who wants the support of an AI that synthesises and spots patterns — never one that decides for you. There are two routes: create a Claude Project (option A, recommended), or export your data and analyse it with whichever AI you prefer (option B).
<aside>
💡
What if I don't use Claude? The instructions on this page and the theoretical context document are plain text: you can paste them into any AI assistant you prefer or that your school has approved. If your assistant can't connect to Notion, copy the entry, paste it into the chat and paste the answer back into the template. And if you'd rather not use AI, the journal works just the same: the structure, the guiding questions and the rubric don't change.
</aside>
Option A — Create a Claude Project from scratch, step by step
A "Project" in Claude is a workspace with its own memory: fixed instructions the AI always follows, and a knowledge folder with documents it can consult. You set it up once, in about 10 minutes, and from then on every conversation inside the project already "knows" how your journal works.
- Create a free account at claude.ai (if you don't have one): email, password, done.
- Create the project: in the left-hand menu, click Projects → + New project. Give it a name (for example, "Teacher's Reflective Journal") and, if you like, a short description.
- Paste the instructions: inside the project you will see an Instructions section ("Set custom instructions"). Copy the text from the "Project instructions" box below and paste it in as is. These instructions apply automatically to every conversation in the project — you never need to repeat them.
- Add the project knowledge: in the project's sidebar there is a knowledge section ("Project knowledge"), a folder where you can upload documents the AI will consult when answering. Upload the "Theoretical context document" from the box below (copy it into a text or Word file and upload it). This way the AI will analyse your entries using the same theoretical basis as the journal (Larrivee's levels, the reflection cycle), not made-up criteria.
- (Optional) Connect Notion: in Settings → Connectors, add the Notion connector and link your account, so Claude can read your journal directly. If you prefer not to connect it, you can also copy and paste your week's entries into the chat — it works just as well.
- Use it: when you want a synthesis, open a chat inside the project and ask: "Synthesise my entries from this week following the project instructions." Review what it returns, correct it if needed, and copy it wherever you want to keep a record (the "Weekly Syntheses" table, your email...).
<aside>
⚠️
With this option, it is you who asks for the synthesis each week — it doesn't happen on its own. Having it run automatically every Monday and leave you a ready-made draft requires a scheduled task, a feature available on Claude's paid plans. It is not needed for the instrument to work: the weekly habit matters more than the automation.
</aside>
- 💬 Project instructions (copy and paste this text)
- 📄 Theoretical context document (copy this into a file and upload it to the project knowledge)
Option B — Export your data and analyse it with the AI you prefer
If you'd rather not connect anything, you can also download your entries and paste them into the chat of any AI (Claude, ChatGPT, Gemini...).
- Open the "Journal Entries" table and filter by the week you are interested in (filter by "Date").
- With that filtered view open, click "•••" (top right of the table) → Export → choose "Markdown & CSV" (to copy the text easily) or "PDF" (to read it). Make sure you export the current, filtered view, not the whole workspace.
- Notion downloads a file: a CSV with one row per entry and one column per field, and/or a text file per entry. Open it with Excel, Google Sheets or a text editor and copy the content of that week's entries.
- Paste that content into the chat of your preferred AI together with this message: