R Roy.Tong

CURRENT CAMPAIGN

Agent Capability Economy

The long-term brand is "how AI becomes products, systems, and markets." The most important structural change right now: software consumers are shifting from humans to agents. That raises a new question — how software is discovered, chosen, invoked, metered, and paid for. This is the only campaign running at full intensity for the next 6–12 months.

THE THESIS

Why now, and why this question

When agents become software consumers, the basic unit of the software economy shifts from the App to the Capability: software is no longer only installed, opened, and browsed by humans, but discovered, chosen, invoked, metered, and paid for by agents.

The shift is already happening, but the industry conversation remains fragmented: some talk about protocols (MCP), some about tools (Skills), some about payments, some about marketplaces, some about observability. What's missing is one complete question connecting it all — how the entire software economy is reconstructed when agents become software consumers.

Within that question, measurement is the gap that must be solved first: without reliable measurement, there is no comparison, no pricing, and no settlement. AgentMeasure is not a product in search of a thesis; it is the productized validation of this judgment.

  • Fragments existSkills · MCP · API · CLI · Agent Runtime · Agent Marketplace · automated transactions · agents using software
  • What's missingOne complete question: how the software economy is reconstructed when agents become software consumers
  • PrerequisiteMeasurement precedes payment: without reliable measurement there is no comparison, pricing, or settlement
  • Category windowA stable discourse has not yet formed — a definition window exists, and this is where authority and citations are built

THREE COLUMNS · ONE CAMPAIGN

Research facts · Essays interpretation · Projects proof

Long-term research stays broad; short-term communication stays narrow. Only this row runs at full intensity; other directions are maintained.

RESEARCHFacts · what is actually happening

Ecosystem map and data ledger

The factual foundation of the agent software ecosystem: capability categories, providers, pricing, distribution, usage evidence, marketplaces, and agent payments. Output is databases, indexes, maps, source ledgers, and monthly updates.

ESSAYSInterpretation · what the facts mean

The body of thought

Abstracting facts into judgments that can be cited and refuted: software consumers shift from humans to agents, the basic unit shifts from App to Capability, and why measurement precedes payment.

PROJECTSProof · can the judgment become something real

Productized validation

Turning judgments into systems that can be tested and contributed to: a measurement layer, an ecosystem map, and an evaluation benchmark.

TERMINOLOGY

Measurement language

Terms are the smallest unit of citation. AgentMeasure is building an open measurement language — every word must be defined clearly first.

  • AgentA software consumer that pursues a user's goal: acquiring context, choosing capabilities, and invoking software to complete tasks.
  • CapabilityA single capability of software that an agent can invoke — not the whole app. The basic unit shifts from App to Capability.
  • ReachWhether a capability opportunity is discovered by agents (discovery layer).
  • ChoiceHow an agent chooses among candidate capabilities (selection layer).
  • UseWhether and how a capability is actually invoked and executed (invocation layer).
  • UtilityHow much the invocation contributed to the task (utility layer).
  • ValueHow utility converts into settlement-ready value (value layer).
  • MeteringReliable measurement of capability use — the prerequisite for pricing and settlement.
  • MarketplaceThe venue where capabilities are discovered, compared, and transacted.
  • CaaSCapability as a Service: software charged by capability rather than by seat or install.

MONTHLY REPORT

Agent Capability Monthly

Build a fixed expectation: people know this will keep updating. One issue per month, fixed structure.

  1. New capabilitiesnew capabilities
  2. New providersnew providers
  3. Pricing changespricing changes
  4. MCP / Skills ecosystemprotocol and tooling ecosystem
  5. Agent commerceagent transactions
  6. Measurement developmentsmeasurement progress
  7. Top 5 signalsthe five most important signals this month
  8. Roy's updated viewRoy's latest judgment (revised with evidence)
Subscribe via RSS ↗ View this month's data foundation

CONTRIBUTE

This campaign is built to be joined

Influence cannot be produced by one person forever. These projects deliberately design interfaces that make contribution easy.

  • Capability IndexAdd a capability: submit a new agent-invocable software
  • AgentMeasureSubmit a measurement case: submit a real measurement
  • BenchmarkAdd your provider: join the ecosystem evaluation
  • Monthly ReportSubmit ecosystem updates: share ecosystem change signals
  • ResearchCorrect an error / add a source
GitHub Issues ↗ AgentMeasure Discussions ↗ Contact Roy