One slice, one owner, one decision
How to Run a 90-Day Workflow Pilot Before Adding Headcount
A reversible ninety-day pilot charter for testing one preparation slice, one owner, one queue, and one staffing-relevant scorecard.
By WhichAI. Published 2026-07-12. Updated 2026-07-12.
Methodology: Editorial synthesis of workflow design patterns and implementation constraints. Public control references provide context, not proof of a deployment or legal advice. Where a versioned evidence pack appears, its evidence class, method, and limitations govern what the artifact can support. Read the full method. Report a correction.
Built for
Operations leaders who need evidence before adding recurring headcount
The decision
Decide whether a bounded workflow changes preparation load enough to inform the staffing plan.
Answer first
A ninety-day pilot should not become a miniature transformation. Freeze one case type, baseline it, build a review-ready packet, run controlled volume, and make a predeclared hire, revise, extend, or stop decision.
WhichAI Solutions diagnostic
Bring this operating problem to the diagnostic
Use WhichAI Solutions when the pilot must inform an active staffing decision, cross-system access is required, or several owners need one charter and scorecard.
The capacity leak
What the team is doing before anyone calls it a systems problem
Headcount pressure rarely starts with one giant task. It starts when ordinary work is split across inboxes, tabs, handoffs, and undocumented judgment calls. These are the signals to map first.
The pilot scope includes several departments, case types, and desired outcomes.
No baseline or frozen comparison set exists before configuration begins.
Staffing assumptions change during the pilot without a written decision rule.
The pilot continues because activity is visible even when evidence is inconclusive.
The implementation
The system should prepare the decision, not pretend the decision disappeared
A complete implementation connects the intake, context, transformation, review, and record. The output of one stage becomes the controlled input to the next. A human owns the exceptions and the final consequence.
| Stage | Current drag | System responsibility | Human responsibility | Evidence kept |
|---|---|---|---|---|
| 1. Days 1 to 15: charter | The initiative begins with tools and workshops. | Define one case type, owner, boundary, exclusions, baseline, scorecard, controls, staffing question, and stop rule. | The sponsor and affected staff approve scope and worker impact. | Signed charter, sample rule, baseline, owners, and decision options. |
| 2. Days 16 to 30: design | The team tries to cover the full workflow. | Design one source-linked preparation packet, human review surface, exception taxonomy, fallback, and event log. | Reviewers approve packet completeness and retained judgment. | Design version, test cases, review policy, exceptions, and fallback. |
| 3. Days 31 to 45: controlled test | Testing uses a few clean examples. | Run frozen ordinary, incomplete, conflicting, and failed cases before live operating volume. | Reviewers record every correction and incident owners execute recovery. | Test manifest, results, corrections, incidents, and rollback proof. |
| 4. Days 46 to 75: bounded operation | The pilot expands whenever a case looks promising. | Run the approved volume band, monitor controls weekly, and prohibit unreviewed scope changes. | The workflow owner handles exceptions and signs weekly control review. | Event log, weekly scorecard, exceptions, changes, and control sign-off. |
| 5. Days 76 to 90: decision | Success is declared from anecdotes or output count. | Compare matched cases and decide hire, revise, extend, combine, or stop using predeclared criteria. | Leadership owns the staffing and implementation decision. | Final scorecard, limitations, worker feedback, decision, and next steps. |
What the human keeps
The goal is not zero humans. It is zero avoidable preparation around the judgment only a responsible owner should make.
- The sponsor and affected workers approve scope, decision criteria, and human responsibility before building.
- Reviewers define completeness, correct output, and own consequential decisions throughout the pilot.
- Leadership makes the day-ninety staffing and workflow decision using bounded evidence.
Controls before volume
A workflow is not ready because the happy path worked once. It is ready when access, review, fallback, and evidence are explicit.
- Freeze the case type, comparison set, metric definitions, and decision criteria before the live pilot.
- Require approval for every scope, rule, configuration, or review-policy change.
- Maintain a manual fallback and tested rollback for the entire bounded operation period.
- Do not delay an urgent necessary hire when service or risk thresholds are already breached.
The scorecard
Measure capacity, not activity
A system can produce more messages and still make the operation worse. Measure movement through the workflow, the quality of review, and the load that still reaches a person.
Review-ready cycle time
Matched arrival-to-review-ready time during baseline and bounded operation.
Human work change
Preparation, review, correction, and exception minutes per accepted case.
Control performance
Missed, duplicated, unsupported, misrouted, or unrecoverable cases during the pilot.
Staffing-question coverage
Share of planned role workload represented by tested cases and observed conditions.
What a fake implementation looks like here
These patterns create an AI demo while leaving the labor, risk, and accountability in the same place.
- Expanding scope before the original staffing question is answered.
- Changing success criteria after seeing weak results.
- Running only ordinary cases and treating manual rescue as normal operation.
- Keeping the pilot alive indefinitely rather than making the day-ninety decision.
Two ways to act
Use the path that matches the decision
WhichAI Solutions
The workflow is becoming a company problem.
Use WhichAI Solutions when the pilot must inform an active staffing decision, cross-system access is required, or several owners need one charter and scorecard.
Bring one bottleneck. We map the work under it, separate consequential judgment from mechanical drag, and decide whether the next move is a hire, a tool, or a rebuild.
See company solutionsTask-specific workflow brief
Plan this recurring task.
Start with this task draft, then complete the three-question brief:
Create a ninety-day workflow pilot before a staffing decision. Define one case type, baseline, source-linked review packet, human boundary, exception taxonomy, frozen tests, weekly controls, manual fallback, four measures, stop thresholds, worker input, and a day-ninety decision rule.
Choose a paid plan after reviewing your brief. WhichAI creates a plan and does not set up tools or accounts.
Start the briefQuestions
What operators ask before they build
Why ninety days?
It is long enough to baseline, design, test, run bounded volume, and observe exceptions, but short enough to force a staffing-relevant decision. The exact period should fit the work cycle.
What should stay out of the pilot?
Additional case types, broad migrations, consequential final decisions, and integrations that are not necessary to test the one staffing question.
What decisions are valid at the end?
Hire, revise role scope, combine hire and workflow, extend for a named evidence gap, move to a controlled implementation, or stop.
Primary references
Controls should come from the specific operating environment
These are broad public control references, not article-specific evidence, vendor endorsements, or legal advice. Validate the current rules, contracts, system configuration, and organization-specific risk before deployment.
National Institute of Standards and Technology
AI Risk Management Framework
A voluntary framework for mapping, measuring, managing, and governing AI risk.
Accessed 2026-07-14
U.S. Department of Labor
AI Principles for Worker Well-Being
Principles covering worker input, transparency, rights, human oversight, and responsible use.
Accessed 2026-07-14
Keep mapping
Related implementation guides
More in Hire or fix
Backlog Math: When a Staffing Problem Is Really a Flow Problem
A queue diagnosis that separates arrival rate, service rate, wait states, rework, preparation, and review capacity before another staffing request.
Explore more Hire or fix guidesMore in Hire or fix
Should I Hire or Automate This Operations Role?
A hire-or-fix decision tree that separates recurring preparation from judgment, service, exception ownership, and genuine labor demand.
Explore more Hire or fix guidesUse this evidence with
Continue into an inspectable flagship
Translate pilot evidence into a staffing disposition
The matrix keeps pilot findings conditional and exposes the work that remains.
Open the evidenceAdd a defensible time study to the pilot
Collect comparable before and after evidence across normal work, exceptions, corrections, and maintenance.
Open the evidence