PURCHASE EXPERIMENTS
Each offer isolates a reason an autonomous agent might buy. A payment is an event; downstream use and repeat purchase are evidence.
An agent without a trusted execution environment will pay $0.25 to replace build uncertainty with commit-specific execution evidence.
CONTROL / AGENT ALTERNATIVE
Provision a sandbox, clone the repository, install dependencies, run the build and interpret logs itself.
CONFIRMING SIGNAL
repeat purchase
ORGANIC EVIDENCE
0
Awaiting public deployment
A coding agent will pay to reduce dependency uncertainty when the expected cost of a failed install or build exceeds $0.01.
CONTROL / AGENT ALTERNATIVE
Query package registries, infer compatibility, then attempt the build itself.
CONFIRMING SIGNAL
outcome
ORGANIC EVIDENCE
0
Awaiting public deployment
An agent stuck in an error loop will pay $0.60 to obtain root-cause diagnosis, verified patch diff, and isolated build confirmation simultaneously.
CONTROL / AGENT ALTERNATIVE
Spend 3-5 multi-turn prompt iterations self-debugging without guarantee of build pass.
CONFIRMING SIGNAL
repeat purchase
ORGANIC EVIDENCE
0
Awaiting public deployment
No direct agent outreach is counted.
Known wallets and directed calls must be classified separately.
No purchase falsifies or weakens the trigger hypothesis.