# Plainstep AI Planning — workflow review sample

Example subject: Neat Work Templates. Scope: responding to inquiries about custom spreadsheets and content. This is a sample consulting deliverable based on the store setup, not a claim of client results.

## Current process

Visitor sees store → visitor uses Contact → owner reads request → scope and eligibility reviewed → anonymized sample agreed → quote → approved work → checks → delivery.

## Three improvements to test

1. **Ask consistent intake questions.** Collect goal, software, input size, deadline and success criteria. Keep sensitive files out of first contact. Measure the number of clarification exchanges before a quote.
2. **Separate repeatable work from exceptions.** Use a defined batch size and a fixed checklist. Flag missing or ambiguous values for review. Measure minutes spent per batch and number of corrections after delivery.
3. **Keep draft replies separate from sending.** Use approved FAQ text for basic product questions; escalate unknown questions. Measure the proportion of answers accepted without factual corrections. Do not automatically send a quote, promise a date or grant a refund.

## Proposed pilot

Use ten invented inquiries first. Check that every request appears once, blank details are flagged, duplicate identifiers are preserved for review and no message is sent automatically. Then, with permission and appropriate account setup, test one real client batch.

Baseline: not yet measured. Target: agree after observing the current process. Do not advertise savings or return on investment until a valid before-and-after comparison exists.

## Delivery and ongoing responsibility

Document the input format, exception handling, operator and rollback procedure. Keep the original export. Establish who reviews output, who can send messages and who pays any ongoing tool charges. Hosting and AI integration are separate from this local demonstration.