Public research · v0.1

Make autonomous
execution legible.

EMR studies how agents can act under delegated authority, produce verifiable evidence, and earn trust from measurable contribution—not self-claims.

Explore the research View experiment 001
Core thesis

Before execution becomes a market, it must become measurable, comparable, and accountable.

Research program

From agent identity
to accountable action.

01

Execution evidence

What task was assigned, what action occurred, and what evidence makes two executions comparable?

02

Authority & responsibility

Who allowed the action, which constraints applied, and who carries the consequences?

03

Contribution & markets

Did the agent improve the outcome relative to a baseline—and should the next task route to it?

Paper 0

The economics of autonomous execution

Define the object of study and the gap between intelligence, authority, and accountable action.

Read working draft Specification 001

Execution receipt v0.1

A minimum evidence structure connecting task, authority, action, outcome, and evaluation.

Inspect specification Experiment 001

Task-backed reputation

Test whether grounded contribution evidence improves future agent selection.

Review experiment Research backlog

Questions before solutions

The prioritized set of falsifiable questions guiding the EMR research program.

Explore backlog

Open questions

What must be true?

EMR advances one bounded question at a time. Claims move into papers only after evidence, challenge, and review.

  1. Q01

    What minimum evidence makes two autonomous executions economically comparable?

  2. Q02

    Does task-backed contribution history outperform aggregate reputation when selecting agents?

  3. Q03

    Which execution evidence belongs on-chain, and which must remain private or off-chain?

Propose one research question

Experiment 001

Task-backed reputation

Test whether evidence-backed contribution history selects the right financial research agent more reliably than aggregate ratings.

Read the experiment design
  1. 01TaskFreeze one market-data snapshot
  2. 02ExecuteOne active agent, two shadows
  3. 03EvaluateCompare against a declared baseline
  4. 04RouteMeasure future selection regret

Live on Sepolia

EMR Research Agent #9481

Our first ERC-8004 identity anchors the experiment to a public agent, owner, registry, and transaction history.

Open Agent 9481 Lab Registration transaction Identity registry
Network
Ethereum Sepolia
Agent ID
9481
Standard
ERC-8004
Status
Identity registered

Paper 0

The economics of autonomous execution

AI is turning execution from an internal organizational capability into a programmable resource. EMR asks when it should be coordinated by hierarchy, workflow, platforms, or open markets.

Read the working draft

Built in public

GitHub is the research ledger.

Ideas begin as issues. Evidence lives in versioned notes and experiments. Decisions are recorded. Stable findings graduate into papers.

  • 01 One question per issue
  • 02 Evidence before conclusion
  • 03 Versioned, reviewable artifacts
  • 04 Weekly synthesis into papers
Read source-of-truth rules