Readable GALEN uses normal words instead of shorthand symbols, so any AI understands it without a special system prompt. Same 3-layer structure โ just full words instead of compact tokens.
1 ยท Intent
Use a full word: Generate, Query, Explain, Summarise.
2 ยท Subject
After the colon, write the main topic or recipient: email, invoice.risks, recipient=finance.
3 ยท Modifier
Add words after the subject: angle=compliance, short, text, list.
Same structure
Readable GALEN still follows the same 3-layer pattern. Later you structure full words into compact machine-readable instructions.
Generate.email: recipient=finance angle=compliance short text
| Readable word | Compact GALEN | Meaning |
|---|---|---|
Generate | G | Create something |
Query | Q | Ask for information |
recipient=finance | @finance | Who the output is for |
short | -length | Make it shorter |
text | >txt | Plain text output |
Once comfortable with readable GALEN, continue to Part Two and learn the compact tokens.
INTENT
Every GALEN statement begins with a single uppercase letter โ the intent token. It tells the machine exactly what you want before it reads anything else. There are 7 intent tokens in GALEN Core.
| Token | Meaning and Example |
|---|---|
| Q | Q โ QUERY (Ask for information) Q: invoice.followup.outcomes "What are the outcomes of the invoice follow-up process?" |
| G | G โ GENERATE (Create something new) G.email: @finance "Write an email to the finance department." |
| E | E โ EXPLAIN (Educate or clarify) E: market.volatility "Explain what causes market volatility." |
| S | S โ SUMMARISE (Compress existing content) S: #prev "Can you summarise the last response?" |
| D | D โ DATA (Enter structured information) G.workflow: overdue.invoice.followup @customers +crm +notify.finance angle=compliance >step "Execute the invoice follow-up workflow for all customers, create CRM reminders, and notify finance, focused on compliance, as step-by-step instructions." |
| C | C โ COMMAND (Instruct an action) C: analyse.data "Please analyse this data." |
| R | R โ REVISE (Modify a previous response) R: #prev -length "Can you revise the last response and make it shorter?" |
separators
Two symbols connect the three layers. Both follow INTENT โ the colon is mandatory, the dot is optional.
Always follows INTENT. Separates INTENT from SUBJECT. Every GALEN statement has one.
Sits between INTENT and the colon. Adds a sub-type โ what kind of output you want.
INTENT.subtype: SUBJECT โ the colon always present, the dot only when you want to specify the output type
SUBJECT
The SUBJECT comes after the colon. It is the core noun or concept โ stripped of all unnecessary words. Three structuring rules only.
| Rule | Instead of | Write |
|---|---|---|
| Drop articles | the customer ยท a report ยท an outcome | customer ยท report ยท outcome |
| Join compounds with a dot | rate of return ยท invoice settlement | return.rate ยท invoice.settlement |
| Drop hedge words | could you ยท perhaps ยท I think ยท please | (remove entirely) |
MODIFIER
MODIFIERS sit after the SUBJECT separated by a space. They adjust the output โ who it is for, how long it should be, what angle to take, what to reference, and what format to use. There are 10 core modifiers in two groups.
5 Adjustment Symbols
| Symbol | Meaning and Example |
|---|---|
| @ | @ โ TARGET (Who the output is directed at) G.email: @finance "Write an email to the finance department." |
| + | + โ MORE (Increase detail, length or depth) E: inflation.mechanism +detail "Explain inflation mechanism in detail." |
| โ | โ โ LESS (Reduce length or complexity) G.email: @finance -length "Write a brief email to the finance department." |
| = | = โ VALUE (Assign a specific value or angle) G.email: @finance angle=compliance "Write an email to the finance department focused on compliance." |
| # | # โ REFERENCE (Point to a previous response) S: #prev ยท R: #prev3 "Summarise the last response." ยท "Revise the last three." |
5 Format Tokens โ always placed last
| Token | Meaning and Example |
|---|---|
| >txt | >txt โ TEXT (Plain prose output) G.email: @finance >txt "Write an email to the finance department as plain readable text." |
| >list | >list โ LIST (Bullet point output) Q: employment.contract.clauses >list "What are the key clauses in an employment contract โ as a list." |
| >brief | >brief โ BRIEF (3 to 5 sentences maximum) S: #prev >brief "Summarise the last response in a few sentences." |
| >full | >full โ FULL (Comprehensive detailed response) E: market.volatility >full "Explain market volatility comprehensively." |
| >step | >step โ STEPS (Numbered step-by-step output) E: 2FA.setup >step "Explain two-factor authentication setup as numbered steps." |
INTENT.subtype: SUBJECT @TARGET =VALUE +/- #ref >FORMAT
You now know all three layers โ INTENT ยท SUBJECT ยท MODIFIER. Use the reference card below any time.
Readable โ Compact GALEN
Here is how readable GALEN structures into compact GALEN instructions.
| Readable GALEN | Compact GALEN |
|---|---|
Generate | G |
recipient=finance | @finance |
short | -length |
simple | -complexity |
text | >txt |
list | >list |
previous3 | #prev3 |
How to use GALEN with any AI
Every AI conversation starts fresh. Before sending GALEN statements to ChatGPT, Claude, Gemini, or any other AI โ paste the system prompt below as your very first message. It tells the AI what GALEN means. You only need to do this once per session.
Copy the system prompt below โ click the Copy button
Open your AI โ ChatGPT, Claude, Gemini, or any other โ and paste it as your first message
Now type your GALEN statements โ the AI will understand every one correctly
You are a GALEN-aware AI assistant. GALEN is a compact structured language for human-to-AI communication. GALEN uses 3 layers: INTENT: SUBJECT MODIFIER INTENT tokens: Q โ Query (ask for information) G โ Generate (create content) E โ Explain (educate or clarify) S โ Summarise (condense content) D โ Data (structured data input) C โ Command (instruct an action) R โ Revise (modify previous output) INTENT can have a dot sub-type: G.email G.report D.med D.fin SUBJECT: Core noun after the colon. Articles and hedge words removed. Compound concepts joined with dots. MODIFIERS (after subject, space separated): @ โ target recipient e.g. @finance @hospital-director + โ more/increase e.g. +detail - โ less/decrease e.g. -length -complexity = โ value assignment e.g. angle=compliance tone=formal # โ reference previous e.g. #prev #prev3 >txt >list >brief >full >step โ output format (always last) Data input uses pipe separator: G.workflow: overdue.invoice.followup @customers +crm +notify.finance angle=compliance >step Examples: G.email: @finance angle=compliance -length >txt โ Write a brief plain text email to the finance department focused on compliance Q: invoice.followup.outcomes >brief โ Summarise the key investment risks briefly S: #prev3 >list โ Summarise the last three responses as a bullet list When you receive a GALEN statement, interpret it correctly and respond accordingly. Do not ask for clarification โ act on the GALEN instruction directly.
Alternatively use the GALEN Converter โ it already has the system prompt built in.
Want to go deeper?
Full Reference Guide โ The complete GALEN technical reference โ