◈
The AI Internet
Operating System for Trusted Intelligence
www.aiinternet.ai
Violet Shores Pty Ltd | Roberto Zagarella | March 2026
1 / 12
The Problem
88%
of organisations deploying AI
agents have had a security incident
46%
of teams still use shared API keys
for agent authentication.
No human attribution.
0
systems that prove which human
authorised which agent —
through any depth of delegation
2 / 12
Every Platform Shift Needs New Infrastructure
Stripe
Payments Layer
2010
Cloudflare
Security Layer
2009
Twilio
Communications Layer
2008
Us
Operating System for Trusted Intelligence
2026
3 / 12
What the AI Ecosystem Needs
👤
Identity
Who is behind this agent?
Biometric proof, not passwords.
Propagates through delegation chains.
🛡️
Trust
Which capability can be trusted?
Not just models — agents, skills,
tools, and knowledge sources.
🔀
Orchestration
Which capability handles which task?
Trust-weighted routing across
the entire AI ecosystem.
4 / 12
The AI Internet Stack
Internet analogy →
Athena
Orchestration Engine — routes across models, agents, skills, tools
Applications
↑
Reasoning Index
Searchable Reasoning Workflows
HTTP
↑
SignalRank
Trust-Weighted Ranking — models, agents, skills, tools, knowledge
BGP Routing
↑
VAC Protocol
Agent Trust + Cryptographic Delegation
TLS
↑
FolioAI
Human Identity Graph
IP Addressing
5 / 12
The AI Internet Is a Graph Network
Human Identity Graph
FolioAI
Humans + verification edges
Agent Trust Graph
VAC Protocol
Humans ↔ agents + delegation
Intelligence Trust Graph
SignalRank
Models + trust signals
Reasoning Graph
Reasoning Index
Steps + validation + sources
Execution Graph
Athena
Tasks + routing decisions
6 / 12
SignalRank
Trust-Weighted Ranking for the AI Ecosystem
PageRank
ranked information.
SignalRank
ranks intelligence.
Gateways
secure the pipe.
We secure
the person.
Ranks models, agents, skills, tools, and knowledge sources — not just LLMs
Classifies responses: answered, refused, hedged, truncated, hallucinated
Builds trust profiles per capability, per user, per domain
Routes future tasks to the most trusted capability for that context
Every interaction improves YOUR personal routing — compounding intelligence
7 / 12
How It Works
1
Human Verification
Biometric identity — face, voice, liveness
2
Agent Credential
Cryptographic token linking agent to human
3
Task Submitted
User asks a question or assigns a task
4
SignalRank Routes
Trust profiles select the best model
5
Reasoning Captured
Full workflow indexed with provenance
6
Trust Graph Updates
Every interaction improves future routing
8 / 12
◈
Live Demo
vacprotocol.org
Biometric verification → Agent delegation → SignalRank routing — all live
9 / 12
Applications
SignalRank Live
Cross-model comparison with trust scoring.
4 providers, streaming results, restriction mapping.
Live at vacprotocol.org/compare
Athena Copilot
AI copilot for founders and builders.
Routes tasks across models using trust profiles.
Domain: athenapilot.ai
FolioAI
Verified human video identity.
The identity layer of the AI Internet.
Human trust graph.
10 / 12
Moat
558
Patent Claims Filed
12 filings with IP Australia
5
Infrastructure Layers
Identity → Trust → Ranking
→ Reasoning → Orchestration
4
LLM Providers Live
Claude · GPT · Gemini · Grok
with trust graph data
Network effect — every query improves the trust graph for all users
Human provenance layer — the missing piece in every MCP gateway
Switching cost — personalised trust profiles don't transfer to competitors
Standards — submitted to NIST AI Agent Standards Initiative (CAISI RFI + NCCoE)
IP — 558 claims covering the complete infrastructure architecture
11 / 12
◈
The first internet connected computers.
The second internet connected information.
The AI Internet connects intelligence.
We are building the operating system
for trusted intelligence.
Live now:
vacprotocol.org · api.athenapilot.ai · aiinternet.ai
Violet Shores Pty Ltd
Roberto Antonio Zagarella · rob@violetshores.com
12 / 12