main.py
README.md
config.jsonPROBLEM
Your chat is full.
You still need the project, but a fresh chat does not know your workspace.
RITA it.
RESULT
Give the next chat the indexes and continue from the first prompt.
YOUR WORKSPACE. YOUR CONTROL.RITA IT.GIVE AI THE MAP.
RITA turns your workspace into compact semantic indexes, so your AI can understand your project from the first prompt — without exploring the entire workspace first.
8SOURCE FAMILIES
Python
Markdown
JSON
Runner
TypeScript
TSX
JavaScript
CSS
8SOURCE FAMILIES
138SUMMARIES
~4 MINBENCHMARK TIME
123.5 KBSUMMARY SIZE
PROBLEM → SOLUTION
Three moments when RITA changes the workflow
RITA solves more than one problem. It prepares your workspace for AI before the conversation begins.
PROBLEM
Returning to an older workspace?
A project with dozens or hundreds of files is hard to re-explain from memory.
RITA it.
RESULT
Get a semantic workspace map and start the next chat already in context.
PROBLEM
Don’t want your working AI exploring the repo?
You want control over what the next AI receives.
RITA it.
RESULT
Give AI the map, not the full workspace.
SAME WORKSPACE. A SMARTER CONVERSATION.
HOW IT WORKS
How RITA works
A simple flow built for real workspaces and fresh AI sessions.
01
DROP
src/
main.py
README.md
...
02
MAP
Code
Docs
APIs
Structure
RITA
03
DOWNLOAD
04
PROMPT
New chat
RITA.zip×
Here's my workspace.
Let's dive in.
SAME WORKSPACE. A SMARTER CONVERSATION.
COMPRESSION PROOF
From workspace to workspace map
RITA compresses raw workspace complexity into a compact AI-readable map.
WORKSPACE
3.8 MB
REAL PROJECT EXAMPLE
Real benchmark project
~103,668 lines
8 source families
Multiple dependencies
RITA
SEMANTIC INDEXES
WORKSPACE MAP
123.5 KB
COMPACT SEMANTIC INDEXES
138 semantic summaries
Source references
Anchors
Portable indexes
≈31× smaller
A real example of how RITA turns a larger workspace into compact semantic indexes.
Semantic summariesAI-ready context
Source referencesTrace back anytime
AnchorsStable, precise context
Portable indexesUse across tools
WHAT YOU GET
What you get from RITA
RITA gives you a workspace map your next AI can understand immediately.
Separate index families for each source typeRITA generates optimized indexes for all supported source families, including Python, Markdown, JSON, Runner, TypeScript, TSX, JavaScript and CSS.
Pysemantic_workspace_index.jsonSample entry (simplified)JSON
1
Use RITA your way
The same RITA product is available in the browser and as a dedicated Mac app.
ACCESS ON THE WEBRITA WebUse RITA directly on m-pathy.ai.
DESKTOP APPRITA for MacDownload RITA as a dedicated desktop app for macOS.
Download for MacSimple pricing. Fast results.
RITA is billed with credits and built to create a workspace map in minutes.
START FROM€5.99100 credits
Get started with 100 credits and map your workspace right away.
BENCHMARK CREDITS57.64
Measured on a real run with about 103,668 source lines, 3.8 MB and 8 source families.
BENCHMARK OUTPUT123.5 KB
138 semantic summaries across 8 source families.
BENCHMARK TIME~4 minutes
Measured across about 103,668 source lines.
Spend a small amount once to avoid spending your first prompts re-explaining the workspace.
Give AI the map.Not the workspace.
RITA helps you control what the next AI session receives.
No code modification
RITA maps the workspace. It does not rewrite your source code.
Controlled handoff
You decide where the workspace map goes next.
No direct repo access for the working AI
Your next chat can work from the map instead of exploring the full repository first.
Processed to create indexes
RITA processes workspace content to build semantic indexes and returns a compact workspace map.
FAQ
Frequently askedquestions
Everything you need to know before you RITA your workspace.
01What is RITA?
RITA is a semantic workspace mapping application for AI-assisted software development. It turns supported sources in a software project into compact, explicit Semantic Workspace Indexes that help an AI orient itself inside the project.
02How do I use RITA?
Use RITA as a web app on m-pathy.ai or download RITA for Mac and use it as a dedicated desktop app. Upload a project folder or ZIP, start the mapping run, download the generated indexes and add them to the AI workflow where you want to continue working.
03What does RITA cost?
RITA uses m-pathy credits. The smallest package is 100 credits for €5.99. In the current real benchmark, a workspace of about 103,668 source lines, 3.8 MB and 8 source families used 57.64 credits. Actual usage varies with source mix, size and required processing.
Buy credits04How long does RITA take?
In the current real benchmark, RITA mapped a workspace of about 103,668 source lines in around 4 minutes. Runtime varies with workspace size, source mix and required processing.
05What do I get from RITA?
You get a semantic map of your workspace. The indexes contain source references, concise prose explanations of mapped files and structural anchors that identify important functions or structures inside those sources. This gives an AI a compact orientation layer it can use from your first prompt.
06How small are RITA files?
RITA can compress the orientation layer dramatically compared with the original workspace. In the current real benchmark, about 3.8 MB of source produced 123.5 KB of semantic summaries — roughly 31× smaller, or about 3.2% of the source size. The exact ratio varies by project.
07Does RITA send my files to an LLM?
Yes. RITA uses an LLM for the semantic summarization stage of the mapping pipeline. Before that step, RITA performs deterministic source admission, fingerprinting, structural anchor extraction and bounded orientation preparation. The LLM is one part of the pipeline — not the entire architecture.
08Does my working AI need access to my repository?
No. You can give your working AI the RITA indexes instead of giving that AI direct access to your complete repository. That is one of RITA’s core use cases: Give AI the map. Not the workspace.
09Does RITA modify my code?
No. RITA maps the workspace and produces semantic index artifacts. It does not automatically edit source files, commit code or push to Git.
10Does RITA support every programming language?
Not yet. RITA currently supports eight source families: Python, Markdown, JSON, Runner / Shell, TypeScript, TSX, JavaScript and CSS.
11Can I use the RITA map with different AI systems?
The RITA output is delivered as explicit JSON index files and is not tied to a single coding agent. You decide which AI or workflow receives the map. Whether a specific external system consumes the indexes directly depends on that system and on how you choose to provide the files.
12Is RITA a coding agent?
No. RITA does not try to replace your AI coding tool. RITA prepares the workspace orientation your AI needs before the work begins.
YOUR WORKSPACE. YOUR CONTROL.RITA IT.GIVE AI THE MAP.
Your next AI session should start with the project — not with hours of rediscovery. Map your workspace once. Take the map with you. Keep building.