In this post I would like to introduce my first thoughts about the general architectural idea behind 'Internet-Fix 1.0'.

The Proxy

In its core 'Internet-Fix 1.0' is a smart proxy server acting in a white-label-mode. This means the proxy accepts only requests from a white listed domain. Domains can be registered by users and voted out by the community. A 'Internet-Fix 1.0 Proxy-Server' is also named 'proxy-space'. A proxy space may be assigned to a general topic or a specific area of knowledge to offer domains from this area in a focused way.

Proxy Network

Proxy Spaces can be connected to each other in the meaning of a trusted proxy-space network. A proxy-space may accept domains from a connected/trusted proxy even if the domain is not known by the proxy space itself.

The Community

The proxy server can be used by everyone. To register a new domain or to vote for a domain user must be a member of the proxy space. There for users can register themselves on the proxy server and become part of the community.

Voting

New domains can be registered easily by every member. Voting a domain out, has to happen through a vote within the community. Domains that get voted out end up on a blacklist and can’t be accessed even through trusted spaces.

To make a proxy space truly useful, it should offer more than just access control – it needs a way to actively discover relevant content. For this reason, each proxy space can offer a search feature based on a lightweight RAG mechanism (Retrieval-Augmented Generation).

Here's how it works in simple terms:

  1. Indexing – Every time a new domain is whitelisted, its text content is automatically split into chunks and converted into vector embeddings using a small, efficient LLM.
  2. Storage – These vectors are stored in a local vector database (e.g., Cassandra, Chroma, FAISS, or SQLite with vector extensions).
  3. Query – When a user enters a search term, the same LLM transforms the query into a vector. The system then performs a similarity search in the vector database.
  4. Results – The most relevant text passages are returned as search results – ranked by semantic similarity, not just keyword matching.

This gives the proxy space a smart full-text search that understands meaning. No external search engine is required, and no user data leaks to third parties.

Optionally, the LLM can also generate a short summary for each result – but that's disabled by default to keep resource usage low.

Open Implementation – API-First & Language-Agnostic

One of the core principles of Internet-Fix 1.0 is freedom of implementation. There is no fixed implementation. Instead, the proxy-space behavior is defined by a minimal, open API contract.

In this way the implementation is completely free – anyone can build a proxy space in any programming language they like. Internet-Fix 1.0 defines an open API contract, not a specific codebase. It's API-first, stack-agnostic, and designed to be as open as the web itself was meant to be