Format: 2 sessions x 2 hours = 4 hours total, across 2 days
Audience: Employees newly issued ChatGPT licenses, first exposure to structured AI training
v4 change: Competency framework replaced with the AI Capability Framework (Augmentation / Automation / Agentic) from AI Enablement Academy.
Learning Outcomes
By the end of this 4-hour program, participants will be able to:
- Explain in plain language what AI/LLMs are, how they generate responses, and why confident output is not always correct.
- Locate themselves on the AI Capability Framework and understand the three distinct layers of AI capability.
- Identify recurring work tasks that are realistic candidates for AI assistance.
- Apply data protection rules correctly, including what can and cannot be entered into ChatGPT.
- Understand tokens and cost drivers well enough to scope prompts efficiently.
- Write effective prompts using a repeatable structure (Context + Task + Format).
- Apply a verification habit to every AI output before using it in real work.
- Leave with at least one live workflow already built and one committed action for the coming week.
What This Package Includes
| Deliverable |
What it gives you |
| 2-session course plan (this document) |
Full minute-by-minute agenda for both days, facilitation notes, and talking points |
| Slide deck outline |
Section-by-section slide content ready to build out in PowerPoint/Google Slides |
| Prompt template library |
Ready-to-use prompts participants can copy into ChatGPT on day one |
| AI Capability Framework |
A 3x4 grid (layer x level) used for self-positioning at the start and end of training |
| Quick-reference handouts |
One-page cheat sheets on data protection, tokens/cost, prompting technique, and output caution |
| Curated learning links |
Vetted, free resources for participants who want to keep building skills after the session |
| Facilitator checklist |
Pre-session prep list so nothing is missed on the day |
The intent is that a participant walks out of Day 2 with working prompt templates already applied to real tasks, not just notes from a lecture.
Design Rationale
- Two 2-hour blocks provide enough contiguous time for foundational concepts and hands-on practice without losing momentum.
- Sequence: concept → data protection → use cases + cost awareness → map to own work → live practical examples with prompting technique → guided practice → action plan.
- The Capability Framework opens Day 1 before any content so participants discover where they sit and how much territory exists beyond it.
- Data protection comes immediately after core concepts, before anyone touches the tool live.
- Cost/token awareness is folded into use-case mapping.
- Caution on AI output quality is introduced at the end of Day 1 and reinforced throughout Day 2.
Day 1 — Foundations, Protection & Use Case Mapping (2 hours)
| Time |
Segment |
Details |
| 0:00–0:15 |
Capability Framework self-positioning |
Show the full 3x4 grid. Participants locate themselves in Augmentation only, typically at L1 Foundations. Highlight untouched Automation and Agentic columns to make scope visible. |
| 0:15–0:40 |
What AI/LLMs actually are |
Plain-language coverage: pattern prediction vs “thinking,” what prompts do, and why confident-sounding output is not always correct. |
| 0:40–1:00 |
Data protection: the ground rules |
Treat every AI chat like a public forum. Cover licensed tier settings, sensitive data boundaries, anonymization, and company policy checks. |
| 1:00–1:20 |
Live use-case demo (before/after) |
Facilitator demonstrates 2–3 common tasks the old way, then with ChatGPT, aligned to attendees’ actual work. |
| 1:20–1:45 |
Workflow mapping + cost-awareness exercise |
Groups list recurring tasks, identify AI candidates, and learn that tighter scope lowers cost and improves output quality. |
| 1:45–2:00 |
Caution on outputs + close |
Hallucination risk: confident tone does not equal correctness. Rule: nothing goes out unchecked. Preview Day 2. |
Day 2 — Prompting, Practical Examples & Guided Practice (2 hours)
| Time |
Segment |
Details |
| 0:00–0:15 |
Recap + framework re-check |
Revisit the grid and frame movement from L1 toward L2 in Augmentation as the realistic near-term target. |
| 0:15–0:30 |
How to prompt for better results |
Teach Context + Task + Format with specificity, decomposition, and experience-level calibration. |
| 0:30–1:05 |
Live practical examples |
Use mapped Day 1 tasks and show weak prompt → refined prompt → usable output, including token-efficiency effects. |
| 1:05–1:45 |
Hands-on practice exercises |
Participants build reusable prompts for real tasks while practicing output verification on every result. |
| 1:45–2:00 |
Share-outs + action plan |
Each participant commits to one AI-backed task for this week and one verification habit; assign an internal AI champion per team. |
Optional follow-up: a 15-minute “what did you try” check-in two weeks later.
The AI Capability Framework
Sourced from AI Enablement Academy’s model of AI capability building. The framework has two axes:
- Enablement layer (horizontal axis): Augmentation (personal productivity), Automation (process layer), Agentic (system innovation).
- Capability level (vertical axis): L1 Foundations, L2 Essentials, L3 Fluency, L4 Native.
| Layer → / Level ↓ |
Augmentation (Personal productivity) |
Automation (Process layer) |
Agentic (System innovation) |
| L1 Foundations |
Identify appropriate AI use cases in daily work and select AI tools that match task requirements. |
Map existing processes to identify automation opportunities with conditional logic and error handling. |
Apply agentic system design principles to architect AI-native solutions with context management and memory. |
| L2 Essentials |
Produce AI assets and artifacts using structured prompting with context control. |
Build AI-enabled interactive tools using no-code platforms. |
Design organizational knowledge systems with MCP integrations, custom skills, and documented decision logic. |
| L3 Fluency |
Design and create custom AI assistants and tools integrated into your stack. |
Create multi-step workflows with agentic nodes and platform-integrated integrations. |
Build production-ready agentic systems with PRD-driven development, testing protocols, and deployment. |
| L4 Native |
Operate with an AI-first personal workflow where automation and agentic delegation are default modes. |
Deploy production-grade automation infrastructure with versioning and handoff protocols. |
Scale agentic systems across teams with governance frameworks and organizational memory architecture. |
How to use this in the training
This 3x4 grid is a stronger “how little you know” tool than a single-track ladder because it makes three separate capability dimensions visible.
For this specific 4-hour training:
- Position participants within Augmentation only. Keep Automation/Agentic visible but explicitly out of scope for this session.
- Target movement from L1 Foundations toward the start of L2 Essentials in Augmentation.
- Use Automation/Agentic columns as “next chapters,” not a comparison device.
Facilitation tip: frame the untouched columns as direction and scale, not a performance gap.
Quick Reference: Data Protection Rules
- Treat every AI conversation like a public forum.
- Confirm your company’s licensed ChatGPT tier and default data/training settings.
- Anonymize content before pasting: remove names, account numbers, and identifiers.
- Never enter client data, employee records, financials, or credentials unless policy explicitly allows it.
- Understand retention controls: data is not kept forever, but not instantly erased either.
Quick Reference: Tokens & Cost, in Plain English
- A token is roughly three-quarters of a word (about 1,000 tokens per 750 words).
- Every prompt incurs both input and output cost, with output typically priced higher.
- Longer conversations cost more because full history is repeatedly processed.
- Concise, well-scoped prompts usually produce cheaper and better outputs.
Quick Reference: How to Prompt Well
Context + Task + Format formula:
- Context: Who you are, the situation, and relevant background.
- Task: The specific thing you want done.
- Format: The desired output structure, tone, and length.
Additional habits: be specific, split large requests into steps, state experience level, and iterate from draft output.
Quick Reference: Caution on AI Outputs
- AI predicts plausible language; it does not verify truth.
- Confident tone is not evidence of correctness.
- Always check numbers, facts, quotes, and client-facing claims.
- Cross-check important claims with a second independent source.
- Training room rule: no AI output goes out unchecked.
Curated Learning Links
Facilitator Checklist
