Intelligent Hyperlinks
Thoughts become things.
That was the original Dreamineering tagline, twenty years ago. Everything on this platform — the feedback loops, the instruments, the standards — exists to shorten the distance between an idea and a thing that works.
The most important invention in that chain is the hyperlink.
Three Pipes
A hyperlink says "go here." A smart contract says "if this condition is met, value flows here." An agent protocol says "discover what's needed, negotiate terms, execute, settle, verify — autonomously."
Three generations of the same invention. Each one pipes something harder to move.
| Generation | Invention | What It Pipes | Transformation |
|---|---|---|---|
| 1. Hyperlink | Tim Berners-Lee, 1989 | Information | Anyone can publish. The web grows. Network effects compound. |
| 2. Smart contract | Nick Szabo, 1994 | Value | Anyone can pipe value. Mycelium grows. Permissionless commerce. |
| 3. Agent protocol | MCP + A2A + VI, 2024-25 | Intent | Anyone can delegate decisions. Agents coordinate. The loop closes itself. |
Each generation solved the trust problem of the previous one:
- Hyperlinks trusted that the document existed at the other end. Often it didn't. Broken links. 404s. Link rot.
- Smart contracts replaced trust with verification. The code is the contract. Immutable. Auditable. But dumb — they wait for someone to call them.
- Agent protocols give smart contracts initiative. Agents discover, negotiate, transact, and verify. The pipe doesn't just carry — it acts.
Why Now
Three things converged in the last twelve months.
Tool access became standard. MCP gave AI models a protocol for using external tools. Before MCP, every integration was bespoke. Now a model can discover and use any tool that speaks the protocol. The hyperlink equivalent: before HTTP, every document network was proprietary.
Agent commerce got a stack. A2A for communication, UCP for checkout, AP2 for payment authorization. Three open protocols that let agents buy and sell without human intermediaries at every step. The smart contract equivalent: before ERC-20, every token was a custom integration.
Intent became verifiable. Verifiable Intent solved the consent gap. When a human taps a card, the tap is the consent. When an agent transacts, there is no tap. VI creates a cryptographic chain from human approval to agent action. Every autonomous transaction traces back to a specific, bounded authorization.
1989: Click a link → information flows
1994: Call a contract → value flows
2025: Express intent → agents negotiate, execute, settle, verify
The Compounding
Each generation makes the previous one more powerful, not obsolete.
Hyperlinks didn't disappear when smart contracts arrived. Smart contracts run on hyperlinks — every dApp is a web page that calls a contract. Agent protocols don't replace smart contracts. Agents compose smart contracts — Programmable Transaction Blocks chain 1,024 operations in a single transaction.
The stack compounds:
| Layer | Protocol | Example |
|---|---|---|
| Pipe information | HTTP + hyperlinks | Agent discovers a service via /.well-known/agent.json |
| Pipe value | Smart contract + blinks | A URL carries a transaction wherever a link can go |
| Pipe intent | A2A + MCP + VI | Agent negotiates terms, verifies human consent, settles on-chain |
A blink is a hyperlink that carries a transaction. An agent protocol is a hyperlink that carries an entire business process. Same invention. Higher bandwidth.
Piping Diagrams
In P&ID nomenclature, pipes connect instruments. The pipe material determines what can flow through it.
| P&ID | A&ID Equivalent | What Flows |
|---|---|---|
| Copper pipe | Hyperlink | Information — documents, pages, APIs |
| Stainless pipe | Smart contract | Value — tokens, assets, proofs |
| Instrumented pipe | Agent protocol | Intent — delegation, negotiation, verification |
Smart contracts are the piping in Agent & Instrument Diagrams. They sit at the Protocols layer of the knowledge stack — sequencing primitives into value streams.
The pipe metaphor holds: you don't build custom piping for every installation. You use standard fittings. The fitting is audited once. Every system that uses it inherits that trust. The Lindy Effect applies — fittings that survive get more trustworthy, not less.
The Test
Every generation passes the same test: can someone who couldn't before, now?
| Before | After | What Changed |
|---|---|---|
| Publishing required a printing press | Anyone with HTML can publish | Hyperlink |
| Commerce required a bank, a lawyer, an escrow agent | Anyone with a wallet can transact | Smart contract |
| Delegation required hiring, training, managing | Anyone with intent can deploy an agent | Agent protocol |
The pattern: a protocol that removes permission from a bottleneck creates a network effect. The network effect creates a platform. The platform creates things that didn't exist before the protocol.
Thoughts become things. The pipe determines how fast.
Context
- Smart Contracts — Technical depth on EVM, Solana, and Sui platforms
- Agent Protocols — The communication, commerce, and verification stack
- Verifiable Intent — Cryptographic proof linking human intent to agent action
- A&ID Template — Pipe = smart contract in the Agent & Instrument nomenclature
- Blinks and Actions — A URL that carries a transaction wherever a link can go
- Essential Algorithm — Algorithms decide the route; protocols enable the handshake; smart contracts execute both
Links
- Tim Berners-Lee — Information Management: A Proposal — The 1989 document that invented the hyperlink as we know it
- Nick Szabo — Smart Contracts: Building Blocks for Digital Markets — The 1996 paper that named the concept
- Verifiable Intent Specification — Open-source specification from Mastercard and Google
Questions
What happens when the cost of delegating a business process approaches zero?
- If each pipe generation made the previous one more powerful, what does generation four pipe — and what trust problem does it solve?
- Hyperlinks created Google. Smart contracts created DeFi. What does the agent protocol generation create that we cannot name yet?
- If standard fittings compound trust over time, does the protocol with the oldest audited fittings always win — or can a structurally safer newcomer leapfrog?
- When agents can negotiate, execute, and verify autonomously, what remains that only a human can do?