Redesign a Real Workflow Around Claude and Pilot It Once End to End

The usual way Claude enters a team is that one person starts pasting things into it and the workflow around them stays exactly as it was. The gains stay invisible because nobody measured the old process, the practice never spreads because it lives in one person's habits, and when a manager asks what the value is, the honest answer is a feeling. Meanwhile the steps that would benefit most, the ones involving synthesis across scattered material, often go untouched because they were never mapped in the first place.

Redesigning a workflow properly means treating the current one as data. You write down every step, who does it, how long it takes, and what the handoff looks like, and only then ask which steps change. Some steps should be augmented, where a person stays in the loop and Claude does part of the work. Some should be redesigned, where the shape of the step changes entirely. Some should be left alone, and saying which ones and why is as much a part of the answer as the changes. This task makes you map, redesign, pilot on one real instance, measure, and then brief a stakeholder in language that states the limitations alongside the gains.

This is a build-and-submit task rather than a guided lab, and it requires no coding. You map a workflow you genuinely own, step by step, recording who does each step, how long it takes, and what gets handed off, and only then decide which steps change. Every step gets marked as augmented, redesigned, or unchanged, and the ones you deliberately leave alone need a justification like any other call.

You then run the redesigned workflow once on a real instance, paste the verbatim Claude output from the step where the most work moved, and measure duration and quality against your baseline. The task asks you to state what a single pilot run cannot tell you, because that honesty is what makes the rest defensible. The document closes with a stakeholder brief written for the person who controls the decision, covering the value, the limitations, the failure mode you would watch for, and the evidence you would want before rolling it out further.

Grading is rubric-based and explainable. Your submission is scored against weighted criteria covering the baseline map, the per-step classification, the transferable prompts, the pilot evidence, and the stakeholder brief, with per-criterion feedback quoted from your document. The pass threshold is 65 percent and you can resubmit. Workflow integration and solution design is a scored domain on the Claude Certified Associate Foundations exam.