Save up to 63%
on your AI costs

Agent Notation is a human-readable shorthand that slashes token costs across every LLM. No GPU. No fine-tuning. Just a better way to write prompts.

63%avg savings

10 minto learn

ZeroGPU required

Playground

EnglishInput:0tokens

Notation

Output appears here...

Output:0tokens

Real prompt categories, measured token savings.

Category	English	Agent Notation	Savings
Code Generation	35t Write a Python function that takes a list of integers, filters even...	`py fn(lst:int[])→int[]: filter even, sort desc`14t	60%
RAG Query	35t Based on the context provided, answer the following question...	`ctx→answer \| !found→'not in ctx' \| !hallucinate`10t	71%
System Prompt	42t You are a helpful customer service assistant for an e-commerce...	`role:cs(ecom) polite prod+ret+refund esc:complex→human`16t	62%
Data Analysis	32t Analyze the sales data in the CSV file and create a report...	`analyze csv.sales->rpt: rev,top5*rev,yoy%`15t	53%
API Integration	38t Create a JavaScript function that calls the Stripe API...	`js fn→stripe: create_cust→create_sub(price_id) catch→log+ret.fail`12t	68%
Average	36.4t	13.4t	63%

No model changes, no fine-tuning, no GPU. Works with every LLM out of the box.

Notation reads like shorthand, not compression artifacts. Teams can review and edit it directly.

Full lexer, parser, and AST. Build tooling, linters, and formatters on top of a real grammar.

Average 63% token reduction across five benchmark categories. Up to 71% on RAG queries.

CLI, VS Code extension, programmatic API. Compress, expand, validate, lint, and format.

Compressed notation expands back to equivalent English. Meaning is preserved, only verbosity is removed.