# Session 1 Handout: Your Job's Top Ten Tasks

<p class="meta">FAST-06 Occupation-Specific AI Role Labs · 7formations · Instructor: Wadood Chaudhary · All examples fictional</p>

## Why this session matters

General AI training tells you what the tools can do. This course starts from what **your job** does: the ten tasks that eat your week. You'll rate each for AI fit and data risk with peers who do the same work, learn the data rules specific to your role, and turn your first two tasks into procedure cards.

## Key concepts

### 1. Start from tasks, not tools

List the ten tasks that take the most hours per week. Use general names ("answer guest pre-arrival questions," "explain monthly variances to clients"), not specific cases. Same-occupation peers help you spot tasks you forgot and merge duplicates.

### 2. Rate fit and risk

| AI fit | Meaning |
| --- | --- |
| **Good** | Clear source, low harm if wrong, easy to check |
| **Use with care** | Useful, but a mistake matters; needs a strong check or approval |
| **No AI** | Involves data that can't go into the tool, or a judgment about a person |

| Data risk | Meaning |
| --- | --- |
| **Low** | Public or internal generic content |
| **Medium** | Internal business data; sanitize first |
| **High** | Personal information, PHI, applicant/employee records, client financials |

High risk usually means "no AI," or a different task: a template with placeholders instead of the real record.

### 3. Role-specific data rules

| Track | Never enter |
| --- | --- |
| HR and recruiting | Applicant resumes, employee records, complaints, medical/leave details |
| Bookkeeping | Client names with account/card numbers, SSNs/EINs, statements |
| Healthcare front office | Any PHI, including name + appointment date |
| Customer service and sales | Customer names with contact, order, or card details; purchased lists |
| Project coordination | Customer confidential drawings and pricing; employee performance notes |
| Office support | Employee DOB, ID or loyalty numbers, home addresses, HR files |

Massachusetts personal information (G.L. c. 93H, 201 CMR 17.00) = name + SSN, driver's license/state ID, or financial account/card number. PHI is protected under HIPAA. HR postings: employers with 25+ Massachusetts employees must include a pay range (G.L. c. 149, §105F, since October 29, 2025).

### 4. The procedure card

| Field | Example (HR: inclusive posting) |
| --- | --- |
| Task and trigger | New role approved for posting |
| Data rule | Skills list and approved pay range only; no applicant data |
| Steps | 1) Get skills list and range from the hiring manager 2) Run prompt 3) Check 4) Hiring manager approves 5) Post |
| Prompt | Five-part prompt with bias-language self-check |
| Check | Pay range exact; no age/gender-coded words; required qualifications job-related |
| Time | By hand 45 min; AI 3 + check 10 = 13 min |

## Worked example: a front-office inventory (excerpt)

| Task | Hrs/wk | Fit | Risk | Reason |
| --- | --- | --- | --- | --- |
| Confirm appointments by phone | 6 | No | High | Patient-specific; stays in practice system |
| Rewrite patient handouts in plain language | 1 | Care | Low | Generic templates; dentist approves |
| Answer common phone questions | 5 | Good | Low | Script from policy sheet |
| Draft insurance appeal letters | 3 | Care | High → Low | Template with placeholders only |
| Summarize chart notes for referrals | 2 | No | High | PHI |

## Checklist: inventory done right

- [ ] Ten tasks with hours per week.
- [ ] Each rated for fit and risk, with a reason.
- [ ] At least one "no AI" considered honestly.
- [ ] My role's never-enter list memorized.
- [ ] Two procedure cards started.

## Do and don't

| Do | Don't |
| --- | --- |
| Name tasks generally | Describe real customers or patients |
| Rate with peers | Rate alone from tool marketing |
| Turn high-risk tasks into templates | Paste records "just once" |
| Write concrete checks | Write "review it" |
| Include the pay range in MA postings (25+ employees) | Let AI invent requirements or ranges |

## Try this at work

1. Track your time for three days and update your top ten.
2. Show your inventory to a colleague in the same role.
3. Pick one good-fit task and run your card on it this week.

## Glossary

| Term | Meaning |
| --- | --- |
| Task inventory | Your top ten tasks by time, rated for fit and risk |
| AI fit | How suitable a task is for AI assistance |
| Data risk | Sensitivity of the data the task involves |
| Procedure card | Steps, prompt, check, time, data rule for one task |
| Placeholder | A marker like [PATIENT] that replaces real data in templates |

## Key takeaways

- Start from your tasks, not the tool.
- Know your role's never-enter list.
- Every procedure needs a concrete check.