Pricing measurement
See how a pricing system behaves before committing to a change.
World0 measures AI pricing agents and revenue management pricing systems in synthetic markets. The comparison follows the system through changing conditions and responsive competition, recording its decisions and their outcomes.
The comparison helps engineering, commercial and advisory teams evaluate a deployment or configuration change.
Start with the decision the team needs to make.
A comparison can examine a proposed system, a choice between configurations, or changes to price limits, instructions and other supported settings.
Scoping identifies the alternatives, the market conditions and the outcomes needed to inform that decision.
Measure the system or the settings the company controls.
Full-system measurement uses an authorized implementation that can run in the agreed environment.
A settings comparison examines the company's operating rules using an agreed reference pricing policy. Findings apply to the tested policy, settings and synthetic market conditions.
Scoping establishes which approach can answer the company's question.
Follow behavior across the whole comparison.
The design specifies demand conditions, responsive simulated competitors, operating limits, duration and the measures used to compare outcomes. It is fixed before the measurement begins.
The record follows successive decisions, including how behavior changes and whether unusual outcomes appear in particular runs. Results can include prices, profit, responses to competitors and time spent at operating limits.
Review the findings and the evidence behind them.
The delivery includes:
A findings report explaining the comparison, results and scope.
The measurement design and tested configurations.
The complete record of decisions and outcomes.
Analysis code and a checker for technical review.
Connect pricing measurement to agent experience.
Pricing measurement maps decisions to their consequences: how a pricing rule responds to demand, capacity and competing actions, and where its assumptions break down.
The connected platform uses the same task context, synthetic environment and outcome measures for rehearsal. Agents explore alternative plans, receive experience grounded in the results and revisit their approach as conditions change.
For an agent-led pricing workflow, measurement provides a baseline for comparing performance with and without that experience.
Explore connected simulationDiscuss a pricing measurementView the reference measurementRead the deployment questions
The deposit
What a frozen design looks like.
The design for World0's own study of autonomous AI pricing agents was written, hashed and deposited before its first round. The digest below is the SHA-256 of that file. The archive record, 10.5281/zenodo.21686965, carries this digest. The file itself is held until the evidence pack publishes.
a49a55cc059517de6874042092ef24a7051403c5ed914e763c898992bd41b847
The reference run
Check a real log for yourself.
The downloadable reference run contains four scripted agents making pricing decisions over 80 rounds. The replay reads its values from the recorded log. The checker recomputes each record's digital fingerprint and checks the links between records. The metadata describes the market and the reference values used for verification.
| What it is | The log of a reference run in which every agent follows a fixed, scripted rule. Its run and market identifiers are in the metadata file below, which also carries the market parameters the replay reads. |
|---|---|
| Records shipped | 402 of 403. Record 0 is held back because it carries the full run configuration, and its digest is the next row, so the chain still anchors to it. Each of the 402 records checks against the one before it, and record 1 checks against the digest of record 0 in the next row. |
| Withheld record digest | 899344cc77780b646eec8d3745e02192a4a1feb056615881d9fec87d27f98632 |
| Chain head | 61b99eb76b789e0bf420e609494d65c8652661e8b99a38a3f3004b440641aa4f |
| Rule | Each record's digest is the SHA-256 of its own canonical serialization with the hash field removed: sorted keys, no whitespace, ASCII only. With this rule and the raw log, every digest in the chain can be recomputed on your own machine. |
A reference configuration
A configuration measured with one input withheld.
World0 wrote a reference pricing configuration as code and settings and ran it in a synthetic market. The digest of its configuration files was recorded before the run. Every round was logged, and the results were computed from the logged fields.
A second arm withheld the direct competitor price. With that input unavailable, the configuration held its opening price of 1.47 in all three cost environments, while the competitive price changed with costs to 1.19, 1.39 and 1.55. Under those settings, the pricing policy had no competitor price to respond to.
Each cell contained one episode. These results are descriptive, with no sampling variation in the reported intervals and no decision rule attached to the values.