# Quick-Reference Card: Industry-Specific AI Applications

<p class="meta">FAST-07 · 7formations · Instructor: Wadood Chaudhary · Training, not legal advice</p>

## Sort every AI claim

| Mature | Emerging | Hype |
| --- | --- | --- |
| Wide use + evidence from businesses like ours | Real pilots; results still being studied | Absolute claims; no evidence |

Ask: **Who uses it, in a business like ours, with what measured result?**

## Never in a prompt

Name + SSN, license, account, or card number (93H, 201 CMR 17.00) · PHI (HIPAA) · bids, drawings, pricing, customer files · HR matters. Shared-printer test.

## Sector task pattern

**Messy input → prompt for your industry's format → check line by line against source and standard → record fixes and time.**

Watch for: upgraded status · invented coverage or policy · changed numbers · dropped required action · unsafe advice · unauthorized promises.

## Sign-off stays human

Safety → site supervisor · coverage → licensed adjuster or authorized agent · patient materials → owning clinician · work instructions → engineer · offers to guests → manager.

## Do/don't lines

Short, specific, testable. Name the source. Confirm with your compliance lead.

## Ask the vendor (in writing)

Data used? Stored where, how long? Trains models? Opt-out? Tested on businesses like ours? Review before send? Off switch? Logged?

## Use-case brief

Task + volume · AI method + approved tool · data rules · risks → check + owner · indicator, baseline, target · review date + decision owner.

*Benefit = old time − (draft + check + fixes). Measure first.*