Prompting
Your creative preferences
Bring a taste profile from an assistant that already knows your work, and pick which model Muse thinks with.
Every model you use starts with no idea who you are. It does not know that you shoot long, that you dislike centred compositions, or that the last four projects were all industrial. Creative preferences are how you tell it once instead of in every prompt. ## Your AI Pick which model Muse thinks with. The choice sets both the model behind its suggestions and which assistant the import flow below is written for. If you already have a preference between Claude and GPT for writing tasks, that preference is the right one here. Muse reads a board and proposes a move in words, so what you are choosing is a writing model rather than an image one. ## Importing your taste Rather than filling in a questionnaire about yourself, you bring a profile from an assistant that has already seen a lot of your work. 1. Copy the import prompt out of the settings panel 2. Paste it into your own ChatGPT or Claude, the account you actually use 3. It writes a profile from what it already knows about you 4. Paste that profile back into Plnty The result is a description of how you work, written by something with months of your conversations behind it. :::tip[Edit it before you save it] Read what came back. Assistants are flattering and occasionally wrong, and a profile saying you favour bold maximalist colour will steer every suggestion you get from then on. Cut the parts that are not true. ::: ## What reads it - **Muse**, when deciding what to suggest - **prompt-enhancer**, when expanding a rough line into a brief Both narrow toward the work you actually make rather than toward a generic centre. ## What it never touches The profile is context, never provenance. It rides along with a request and it is deliberately kept out of everything that gets stored: | Never written to | Which means | |---|---| | An asset's prompt record | Your provenance stays the words you typed | | Workflow data | The board document holds no profile | | Generation logs | Nothing about you accumulates in the run history | | Error reports | A bug report carries the failure, not your profile | :::caution[It is a preference, not a style lock] Preferences steer suggestions and prompt expansions. They do not constrain the tools, and a tool with strong opinions of its own, like **antagonist**, will still argue with you. That is the tool working as intended. ::: ## Skipping it entirely Nothing requires this. Muse works without a profile and the enhancer expands prompts without one. Setting it up narrows both toward you, so it pays off if you lean on either of them regularly and does nothing if you do not.Every model you use starts with no idea who you are. It does not know that you shoot long, that you dislike centred compositions, or that the last four projects were all industrial. Creative preferences are how you tell it once instead of in every prompt.
Your AI
Section titled “Your AI”Pick which model Muse thinks with. The choice sets both the model behind its suggestions and which assistant the import flow below is written for.
If you already have a preference between Claude and GPT for writing tasks, that preference is the right one here. Muse reads a board and proposes a move in words, so what you are choosing is a writing model rather than an image one.
Importing your taste
Section titled “Importing your taste”Rather than filling in a questionnaire about yourself, you bring a profile from an assistant that has already seen a lot of your work.
- Copy the import prompt out of the settings panel
- Paste it into your own ChatGPT or Claude, the account you actually use
- It writes a profile from what it already knows about you
- Paste that profile back into Plnty
The result is a description of how you work, written by something with months of your conversations behind it.
What reads it
Section titled “What reads it”- Muse, when deciding what to suggest
- prompt-enhancer, when expanding a rough line into a brief
Both narrow toward the work you actually make rather than toward a generic centre.
What it never touches
Section titled “What it never touches”The profile is context, never provenance. It rides along with a request and it is deliberately kept out of everything that gets stored:
| Never written to | Which means |
|---|---|
| An asset’s prompt record | Your provenance stays the words you typed |
| Workflow data | The board document holds no profile |
| Generation logs | Nothing about you accumulates in the run history |
| Error reports | A bug report carries the failure, not your profile |
Skipping it entirely
Section titled “Skipping it entirely”Nothing requires this. Muse works without a profile and the enhancer expands prompts without one. Setting it up narrows both toward you, so it pays off if you lean on either of them regularly and does nothing if you do not.