How It Works

Your Private Knowledge Engine

Privacy by architecture. Here's how.

Smart Routing
Internal AI first, external providers only when needed
Context-Aware
Keeps what matters, protects the rest, never logs your content
Private by Design
Your data can't be read without your key

The Two Problems We're Solving

Problem 1: You Can't Control What Generic AI Remembers

You: "What should I get Liam for his birthday?"

ChatGPT: "I recall you have a friend named Liam, but I don't have details on his interests or your budget. Could you remind me?"

❌ Inconsistent memory. You don't know what it uses, keeps, or forgets.

Problem 2: Most AI Tools Send Everything to External Providers

You: "What did my doctor say about my knee?" [You attach your full medical history PDF]

AI: "Based on your records..." [Your entire medical file is sent to an external provider]

❌ You have to send your full private data to external AI just to get a simple answer

THE WHEEL's Solution: Knowledge Compounding + Smart Routing

Internal AI (No External Providers Needed)

You: "What did my doctor say about my knee?"

THE WHEEL: "Your doctor recommended rest for 2 weeks, ice 3x daily, and follow-up if pain persists. Notes from January 15, 2026 appointment."

🔒 THE WHEEL's AI retrieved from YOUR encrypted notes

✓ Instant, private, accurate - not sent to external AI providers

External AI Providers (When Actually Needed)

You: "What should I get Liam for his birthday?"

THE WHEEL: "Based on your notes, Liam mentioned loving vintage watches last time you talked, and your budget shows $200 for his May 15th birthday. A vintage Seiko or Citizen from the 1970s would fit perfectly — great craftsmanship in your budget range."

🔒 THE WHEEL retrieved context → Anonymized → Routed to a fully featured LLM from top AI providers → Answer

✓ Context from YOUR notes + creative reasoning from external AI providers (identity anonymized)

Try it yourself — request beta access →

The three questions we answer:

  1. How do we remember your context without making you repeat yourself?
  2. How do we know when to use THE WHEEL's internal AI vs. when to route to external AI providers?
  3. How do we protect your identity when external AI providers ARE needed?

Save the heavy lifting for when it matters

Many of your questions are about you and your own information: your notes, your documents, your conversations. Those don't need a massive AI model. They need fast, accurate retrieval from your own data. Our internal AI handles that privately. External AI only steps in when you need deeper reasoning or creative help.

1

You ask a question

"What should I get Liam for his birthday?"

2

We search your context — or admit when we can't help

Your notes, conversations, and documents are all encrypted. We search them for relevant context. But if we don't have details about Liam's interests or past gift ideas, we tell you immediately and ask you to clarify. No wasted tokens, no wasted research, no made-up answers.

3

Simple questions stay internal

Looking up something you saved? You get an answer from your notes directly — powered by focused AI models hosted entirely on our infrastructure. Need outside expertise? Your context is routed to a frontier model, but first your identity is protected.

4

Your identity is protected, your question isn't

When external AI is needed, we strip out anything that identifies you while keeping the information needed to actually help. The result: useful answers without exposing who you are.

Your context
"Liam mentioned loving vintage watches last time you talked. His birthday is May 15th, and your budget shows $200 allocated for his gift. He appreciates craftsmanship and unique pieces with history."
What external AI sees
"Looking for a gift for someone interested in vintage timepieces. Budget around $200. Values craftsmanship and historical pieces. Timeline: upcoming celebration."
💡
External AI gets enough context to suggest thoughtful gift ideas. It doesn't see Liam's name, specific dates, or identifying details. Anonymized queries are processed ephemerally — nothing is stored after your answer is delivered. Learn more about our approach on the Technology page →

You get useful, private answers

The response comes back with a thoughtful gift suggestion — a vintage watch that fits your budget. External AI never learned who Liam is, who you are, or the exact date of the birthday. We put the personal details back in for you, so the answer reads naturally.

Bottom line: You get the benefits of powerful AI without giving up your privacy. We protect who you are while letting AI understand what you need help with.

Want to understand the technology behind this?

Learn about the cryptographic foundations, smart routing algorithms, and patent-pending anonymization that makes THE WHEEL possible.

Explore the Technology →

Your knowledge, actually private.

Privacy isn't a feature. It's a fundamental right. We built THE WHEEL so we don't have access to your data without you.

What to expect from the beta?