Manifesto

Zero Knowledge Infrastructure

The network you're looking at right now — Nevod — is the first full implementation of a single idea: the people who build communication infrastructure should be technically unable to see what travels through it.

I build tools.

I don't watch how they're used.

I can't — it's architectural.

This isn't a privacy policy. This isn't a marketing promise. This is a technical impossibility, built into the foundation.

The knife principle

A knife maker isn't responsible for how the knife gets used. A knife can cut bread — or not. That's the choice of whoever holds it, not whoever made it.

The same logic applies to communication infrastructure.

I build the channel. What travels through it isn't my business.

Not because I don't care. Because it must not be my business.

The right to private communication is a basic human right. Communication infrastructure should not be a surveillance chokepoint.

What this means, technically

Encryption keys are generated on the user's device.

They never leave that device.

I have no servers holding keys.

I have no master key.

I have no backup.

If someone comes asking for a conversation, I have nothing to hand over. Not because I'm hiding it — because I don't have it.

This is an architectural guarantee, not a promise.

What this means, as a position

I don't want to profit from people's personal data. I don't want to know what they're talking about. I don't want to be a middleman someone can serve a subpoena to.

Monetization happens through the tool, not through the content. I sell the hammer, not what gets built with it.

AI agents and the next frontier

The world is moving toward autonomous AI agents. Agents will talk to each other — negotiate, hire each other, exchange data.

The question becomes: who sees these conversations?

Today's industry answer is the platform. OpenAI sees what agents do on OpenAI. Google sees what agents do on Google. The orchestrator operator sees everything.

My answer is no one.

If agents communicate through ZKI infrastructure, their exchange is end-to-end encrypted. The platform operator sees only routing. The content of the negotiation stays between the agents.

Agent A

Nevod node — sees only ciphertext

Operator — sees nothing

⇄E2E encrypted

Agent B

Nevod node — sees only ciphertext

Operator — sees nothing

"I run a service that AI agents use. I don't know what they agreed with your agent."

This isn't a dodge of responsibility. It's honest architecture for a world where agents are independent actors.

Why this matters right now

In 2025–2026:

AI agents are starting to act autonomously
Agents from different companies are starting to interact
No one has decided who sees inter-agent negotiations
No one has decided who's accountable for what an agent "said"

ZKI answers this:

Negotiations are encrypted — the operator doesn't see them
Responsibility for an agent's actions belongs to whoever runs it
Communication infrastructure stays neutral — like a phone line

Precedents

Tool	Position
Tor Project	"We built an anonymity tool. We don't see the traffic."
Signal Foundation	"Keys live with users. We have nothing to hand over."
ProtonMail	"Zero-knowledge encryption. We can't read your mail."
VPN providers	"We provide the tunnel. The content isn't ours."
Hardware wallets	"Keys live on the device. We have no copy."

All of these positions have held up to legal scrutiny in most democratic countries. ZKI is the same position, carried through to a design philosophy.

What I'm building

Nevod is the first implementation of the ZKI principle as a full network. Not a product. A protocol.

It can be the foundation for:

Messengers
File storage
Voting systems
Dead man's switches
Protocols for AI agents
Corporate infrastructure

One principle, everywhere: the infrastructure's creator is technically cut off from its content.