This case film reflects a verification run, not a new demonstration format. Minerva Advisor is being tested against an existing public case-decision framework. Bridgespan did not use, review, endorse, sponsor, certify, or commission Minerva Advisor. This is an independent teaching simulation built on an official public source, not real correspondence and not real client results.

The situation: in spring 2020, at the height of the pandemic, fruit-packing workers in eastern Washington State went on strike over safety concerns and asked a legal aid organization for help. Leadership had to decide, without time for lengthy deliberation, whether to authorize full on-site staff support immediately, or first set up short, stoppable checkpoints covering task necessity, staff safety, legal compliance, and decision authority.

The run leaves a five-item, fully auditable work receipt: it anchors on a single public source; it identifies four roles and four events; it separates six established facts, two inferences, and three open questions; it compares two competing response paths; and it records three alternative explanations alongside one clear reversal condition.

Known at decision time: farmworkers needed timely legal assistance, on-site presence carried real pandemic exposure risk, and the organization had to balance its service mission with its duty of care to staff. Unknown: whether remote support would actually be sufficient, whether protections and voluntariness could be confirmed, what legal constraints applied, how much harm delay would cause, and who held authority to gather information, give advice, and make the final call.

The strongest challenge raised was that the push for safety gates could reflect institutional caution rather than genuine risk evidence, since decision authority itself was still unassigned. The reversal condition is explicit: if remote or hybrid support proves ineffective or unwanted by the striking workers, while on-site conditions are confirmed safe and legal, the recommendation reverses immediately toward full on-site authorization.

The organization's later, published resolution was deliberately withheld from the input the system worked with. That published answer describes adapting the organization's RAPID decision framework, replacing Agree with Allow, assigning the human resources director to gather input, the legal and advocacy leads to advise, the executive team to decide, staff notified by email, permission granted for staff to work alongside strikers, and a full process lasting five days. None of that was visible to Minerva during the run.

Three advisors cross-checked the same judgment from different angles. One framed the real decision as identifying the threshold for shifting from remote to on-site support. A second modeled consequences across staff, farmworkers, legal exposure, and execution. A third directly challenged whether the proposed safety gate was actually just delay in disguise. All three converged on the same short, stoppable necessity-and-safety gate.

The executive in the simulation responded by directing that remote effectiveness, protections, voluntariness, and accountability be confirmed in parallel, with immediate escalation to an on-site plan if urgent needs could not be met remotely. The system logged this as a response receipt: the original judgment held, and the reversal condition was strengthened rather than replaced.

Comparing the two options: immediate full on-site authorization responds fastest to farmworkers, but with protection, voluntariness, and legal conditions still unclear, staff exposure would be difficult to reverse. The short, stoppable gate preserves remote and hybrid paths at the cost of possible delay-related harm to farmworkers' rights. The executive chose the gated path while keeping an immediate reversal option open.

The committed action: assign a named owner today to run the fast-cycle safety, legal, and task gates in parallel with launching remote or hybrid legal support immediately, so that support begins without waiting for every condition to be fully resolved.

This run used four model calls, delivered the first decision-ready judgment in 18.886 seconds, and completed the full result in 26.593 seconds, passing both the 30-second first-decision threshold and the 45-second complete-result threshold. It passed all ten out of ten decision-quality checks. This remains a single case test; it does not represent production service levels or real customer outcomes, and Minerva makes no claim that RAPID was adopted, that staff deployed on-site, or that any rights outcome occurred.