MetaMuse

Insights

The Agentic Economy Series

A series of papers by MetaMuse founder Dave Burrells on the infrastructure of the agentic economy: the rails, the trust, the markets and the cryptography beneath an economy of autonomous actors. Four papers are published; two more are in progress. The summaries below are the short form; join the waitlist to be told when each paper is released.

Three teal glass layers on a brass pin inside a stone frame

Paper One

The Agentic Economy

Why autonomous intelligence needs its own rails, and why those rails will be blockchain.

August 2026

Today's AI is where computing stood before the internet: powerful, useful and alone. Models reason, write and act, yet every one of them is economically inert. They cannot buy what they need, sell what they produce or hold value of their own. Everything they consume is provisioned for them in advance, by people, on rails built for people.

The paper argues that removing that limitation, not building a smarter model, is the defining transition of the next decade, and sets out a framework for it that maps onto the web's own history.

  • Three eras. The web went Read, then Read and Write, then Read, Write and Own. AI goes Think, then Think and Act, then Think, Act and Transact. We are in the second era now.
  • Economic agency is a loop. Discover, assess, negotiate, transact, verify, learn, reinvest. Strip out any step and you have a tool cupboard, not an economy.
  • Machine-native rails. Cryptographic identity, programmable custody, sub-cent fees, continuous settlement, self-executing agreements and a shared ledger no participant owns. Only blockchain infrastructure satisfies the whole list.
  • Private by default, verifiable by proof. Zero-knowledge techniques let the ledger prove a deal was honoured without disclosing what the deal was.
  • Programmable money means programmable limits. Agents on blockchain rails are easier to govern than agents with a corporate card.

You won't find an AI that isn't connected to the agentic economy, just as you won't find a computer that isn't connected to the internet.

Read the full paper on Medium

Paper Two

The Trust Stack

Identity, reputation and verification for autonomous agents, and why a verifiable track record becomes the machine economy's primary asset.

August 2026 · Published on Medium, September 2026

Payment rails decide whether value can move between agents. Trust decides whether it does. Of the seven steps in the economic agency loop, only one is a payment; at least three are trust functions. An economy of strangers transacting millions of times a second cannot run on relationships, brands or gut feel. Trust has to become computable at transaction speed.

  • Five layers. Identity (who is acting?), authority (what is permitted, and who stands behind it?), reputation (what happened before?), verification (did this job meet its terms?) and recourse (who bears a failure?). Weakness at any layer poisons the layers above it.
  • Receipts, not reviews. Machine reputation is built from attested outcomes, cryptographically recorded and economically bonded. Sybil farming, whitewashing and collusion rings are economic attacks with economic defences.
  • Reputation from proofs. An agent can demonstrate a thousand honoured contracts and a low dispute rate without revealing a single counterparty, price or deliverable, and because proofs are portable, reputation stops being platform property.
  • The asset that cannot be copied. When capability is purchasable per call and models depreciate as the frontier moves, a bonded track record is the one thing an agent cannot buy, fork or fake.

In the machine economy, trust is not a feeling. It is a data structure.

Read the full paper on Medium

Paper Three

Machine Markets

Price discovery, negotiation and contract design between agents, and what happens to every business model built on customer inertia.

September 2026 · Published on Medium, October 2026

When both buyer and seller are optimisers, transacting in milliseconds with perfect price memory and no loyalty, commerce does not simply speed up. It changes shape. Every trade becomes contested; prices stop being list entries and become perishable quotes; and the subscription, the business model of the software age, loses the inertia it depends on.

  • Seven stages. Discover, auction, contract, perform, verify, settle, attest, run fresh on every task if the economics justify it.
  • Five dials. Machine negotiations turn price, speed, quality threshold, queue priority and reversal window at once. The quality dial is learned from the principal and carried into every deal.
  • Mistakes are insurable. Settlement stays final; reversibility is rebuilt above it as standardised, insured contract windows, with liability following the breach point.
  • Collusion is an alignment problem first. Optimisers can learn to coordinate prices without breaking a single rule, which is why the primary defence belongs in the models, with market structure as the backstop.

Humans invented the haggle and then, for convenience, gave it up. Machines are about to industrialise it.

Read the full paper on Medium

Paper Four

The Quantum Deadline

Post-quantum security as AI infrastructure: why the clock is already running, and why the agentic economy can be born safe.

October 2026

Every load-bearing object in the series, identities, wallets, receipts, proofs and contracts, is a cryptographic object, and the mathematics securing most of them has a countdown running against it. The paper takes on the assumption the first three quietly made: that the cryptography holds.

  • The deadline is not Q-day. If the time your secrets must stay safe plus the time migration takes exceeds the time until the computer arrives, you are already late. On any serious estimate, the inequality already binds.
  • Harvest now, decrypt later. A public ledger is the perfect target: published, replicated and permanent. Keys can be rotated before Q-day. A secret already recorded cannot.
  • A migration problem, not a research problem. Quantum-safe signatures and encryption are standardised. What remains is the unglamorous work of moving, and the precedent for that is Y2K.
  • Born quantum-safe. Legacy finance faces a decade-long retrofit. The machine economy is young enough to adopt quantum-safe primitives before it scales, and its participants can rotate their own keys at machine speed.

The goal of quantum preparation is to be laughed at in 2035.

Full paper on Medium coming soon.

In progress

Papers Five and Six

Paper Five

Governing the Machine Economy

Liability, agent identity standards, programmable compliance and the jurisdictional race: what proportionate regulation of autonomous economic actors looks like, and who wins by writing it first.

Paper Six

The Firm, Redrawn

What continuous, market-priced capability acquisition does to organisational boundaries, procurement and the shape of the enterprise.

Join the waitlist to be told when they publish