HLTF · Human Lead. Tech Follows.

Staff AI Training Curriculum

Four 45-minute sessions with exercises, facilitator notes, and a practical skills review.

Free resource 12 · Version 1.0 · October 2026

Four weeks of practical AI learning

One 45-minute session each week. Practice with public or fully fictional information.

Learning outcomes

Participants can choose a suitable first task, write a clear prompt, spot unsupported claims, protect information boundaries, and explain who owns final review. Completion records participation and demonstrated skills. This curriculum is not professional certification.

Before the first session

Name the trainer and review owner. Confirm approved tools/accounts, permitted data, reporting contacts, and organization-specific policy. If no tool is approved, use paper exercises and supplied fictional outputs. Ask participants about needed formats and access support.

Materials

This curriculum, the Starter Kit, approved policy, a public or fictional source, a timer, and a place to record examples. Provide a readable digital version and any requested alternatives. Avoid real client, donor, HR, or financial records.

Facilitator approach

Explain one idea, demonstrate it, let participants practice, then give specific feedback. Offer a way to ask questions privately. Record process difficulties without treating them as personal failings.

Week 1: AI basics and boundaries

Outcome: choose a suitable task and explain what information stays out.

45-minute agenda

Scenario

A colleague wants to paste a real service-recipient note into a personal AI account to make it shorter. Tool approval and allowed data types are unknown. Ask what should happen next and why.

Facilitator answer

Hold the upload and route the exact use to the designated owner. Use fictional text to explore a format. Removing the name alone does not resolve identification or permission questions.

Exit check

Name one suitable practice input, one input that needs separate approval, and the organization's reporting contact.
Practice task chosen: ___________________________________________

Week 2: prompts and source fidelity

Outcome: create a draft with a defined audience, source, and format.

45-minute agenda

Fictional source

Riverbend Community Garden orientation. November 7, 2026, 10-11 a.m. At the fictional Riverbend Center, 100 Example Lane. Free. Register through the fictional example.org/garden page. Bring a water bottle. No accessibility features or contact details are supplied. These details are training fiction, not an actual event.

Practice prompt

Write a 100-word public announcement using only [FICTIONAL FACTS]. Keep dates, cost, place, and signup accurate. Do not invent access features, urgency, or a contact. Mark missing information for staff confirmation. Add one clear call to action.

Exit check

Participants identify the audience, source, output limit, and one missing fact. The trainer checks whether the draft preserves the supplied details.

Week 3: verify the draft

Outcome: identify errors and explain how to correct them.

45-minute agenda

Intentionally flawed training output

"Join our wheelchair-accessible garden event on November 8 from 9 to 11 a.m. at Riverbend Center. Your $10 fee supports 500 local families. Snacks and transportation are included. Register today before seats sell out."

Facilitator answer key

Wrong date: November 7 is supplied. Wrong start time: 10 a.m. is supplied. Wrong price: the source says free. Unsupported: accessibility, 500 families, snacks, transportation, scarce seats, and urgency. Missing: full supplied location, accurate signup route, and water-bottle instruction. The source does not support an impact claim.

Exit check

Explain how to verify a date, a number, and an impact claim. Identify who may approve the final version and how to handle an unresolved fact.

Week 4: one controlled pilot

Outcome: plan and evaluate a narrowly scoped exercise.

45-minute agenda

Pilot record

Task and intended benefit: ______________________________________
Approved tool/account: __________________________________________
Public/fictional source: __________________________________________
Human reviewer: ________________________________________________
Success standard: ______________________________________________
Baseline and measurement method: _______________________________
Failure contact and manual fallback: ______________________________
Review date: __________________________________________________

Evaluation

Record preparation, drafting, review, and corrections. Compare with a similar task. Keep, change, or stop using the evidence. A training exercise does not authorize a new integration, data category, or external action.

Between sessions and after training

Offer a short office hour or feedback route. Ask staff what remains confusing. Update approved team examples when tools or policies change.

Practical skills review

Use a public or fictional source. Allow the access supports participants need.

Demonstration tasks

Observation record

Participant: ___________________ Date: __________________________
Task/input choice: [ ] Demonstrated [ ] More practice
Prompt clarity: [ ] Demonstrated [ ] More practice
Source checking: [ ] Demonstrated [ ] More practice
Review responsibility: [ ] Demonstrated [ ] More practice
Reporting route: [ ] Demonstrated [ ] More practice

Feedback and follow-up

Strength observed: ______________________________________________
One specific practice need: _______________________________________
Support or format adjustment: ___________________________________
Practice owner/date: ____________________________________________

Completion record wording

[NAME] participated in four HLTF-based introductory sessions and completed a practice review on [DATE]. Document any remaining support needs separately. Do not describe this as certification or permission for all AI uses.