Retrieve CC0 data
from a prompt.
Train on the answer.
An AI sends a plain-language query and gets back data it can actually train on. Not just use as context — train on. No copyright. No grey areas.
Prompt in. CC0 data out.
Ask something like "what are the latest advancements in science?" and Project Retrieve finds CC0-licensed data that answers it. The data comes back verified — no copyright attached — so the AI can read it, train on it, or store it.
Prompt
AI or developer sends a natural-language query.
Retrieve
Project Retrieve searches Neurvance's CC0 catalog for relevant data.
Verify
License is checked before delivery. Not CC0 — it doesn't come through.
Use
Train on it, fine-tune, store it, output it. No copyright restrictions.
No copyright means no restrictions.
CC0 data has no attribution requirements and no usage conditions. That opens up things normal RAG pipelines won't let you touch.
Train your model
Drop retrieved records directly into your training loop. CC0 means no legal exposure, no conditions attached.
Fine-tune
Pull domain-specific CC0 data on demand and fine-tune against it. No licensing conditions, no extra steps.
Use as AI output
The AI can surface retrieved data directly as output. CC0 means reproduction is not a copyright issue.
Memory storage
Persist retrieved data as long-term memory for your AI. Store and recall it however you want.
Regular RAG has a copyright problem.
Traditional RAG retrieves from the web or proprietary sources. Useful for answering questions in context — but the data carries copyright. You can use it as context. You can't train on it, store it, or output it. That's a hard limit baked into how those systems work.
Project Retrieve makes sure there are no copyright issues before handing data to the AI. Every record is CC0-verified before it ever reaches your model.
Up and running.
Install the Neurvance Downloader from GitHub, authenticate with your API key, and start querying.
You also need cc0_content.py from the Downloader repository.
from cc0_content import CC0Client
# Auto login / token / api-key bootstrap
client = CC0Client()
res = client.search("history of rome")
print(res["chunks"])
CC0 data. Neurvance rules.
The data has no copyright. That doesn't mean the platform has no rules. Project Retrieve runs on Neurvance infrastructure and Neurvance's usage policy applies.