AI Accelerator Learning Hub

Delivered in partnership with Women in Digital + Ai Her Way

Welcome to your central hub for everything throughout the program. Here you'll find session recordings, resources, tools, and links for each week as they become available. Bookmark this page and check back after each session.

Define artificial intelligence, machine learning, and large language models
Understand the current AI landscape: tools, capabilities, and limitations
Introduce the AI Maturity Index (5 stages from literacy to agentic AI)
Identify where participants currently sit on the maturity curve
Recognise the difference between helpful AI hype and practical reality

What is AI? (Demystifying the terminology)
How large language models work (without the technical jargon)
The AI tool ecosystem: ChatGPT, Claude, Copilot, Gemini, and specialised tools
Understanding tokens, context windows, and why they matter
The five stages of AI maturity: Literacy → Prompting → Custom Bots → Automations → Agentic AI
Ethics, bias, and governance: Why we can't ignore these questions
The EquiAI Framework: Human oversight, transparency, and measurability
🔒 Governance spotlight: What happens to your data when you use AI tools? Understanding data handling policies across ChatGPT, Copilot, Claude, and Gemini. What's safe to share, what isn't, and how to check your organisation's stance. Practical guidance for participants in regulated industries (health, finance, government).

Pre-program survey results: Share back the cohort snapshot (who's in the room, experience levels, top concerns)
Reflection: Where do you sit on the AI Maturity Index?
First prompt exercise: Testing different AI tools with the same query
Identify 3-5 tasks in your current role that feel time-consuming or repetitive
Quick audit: Does your organisation have an AI use policy? What does it allow/restrict?

Track your time on one repetitive task this week (time audit baseline)
Join the community channel and introduce yourself
Reading: The EquiAI Framework (provided PDF)
Find out: Does your organisation have an AI policy or acceptable use guidelines? Bring what you find to Week 2

Apply the ROSE-T prompting method for consistent, high-quality outputs
Understand prompt engineering fundamentals
Recognise when to use different prompting strategies (zero-shot, few-shot, chain-of-thought)
Iterate and refine prompts based on outputs
Avoid common prompting pitfalls

The ROSE-T Method breakdown:
- Role: Setting context and persona
- Objective: Clear, specific goals
- Specifics: Tone, format, and audience considerations
- Explicits: Providing models of desired output, rules, do and donts
Advanced techniques: Chain-of-thought prompting, prompt chaining, constraining outputs
Working with context: What to include, what to leave out
Creating reusable prompt templates
Understanding AI limitations: hallucinations, outdated information, and bias

Live demonstration: Transforming a vague prompt into a ROSE-T structured request
Paired exercise: Participants share “before and after” prompts
Use case exploration: Writing emails, meeting summaries, content drafting, data analysis

Create 5 ROSE-T prompts for tasks you identified in Week 1
Test each prompt 3 times and refine based on outputs
Document your results: What worked? What didn’t?
Save your best prompts in a personal “prompt library”
Re-time one of your tracked tasks using AI assistance

Create custom GPTs tailored to specific roles or tasks
Structure knowledge bases and reference materials for AI
Design effective instructions and conversation starters
Test and iterate on custom bot behaviour
Understand when custom bots are more appropriate than general-purpose AI

What are custom GPTs, and how do they differ from standard LLMs?
The anatomy of a custom bot: Instructions, conversation starters, knowledge files, capabilities
Creating a knowledge base: What documents to include, how to structure them
Writing system instructions that shape bot behaviour (regardless of platform)
Use cases: Personal assistants, role-specific advisors, company knowledge bots
Cloning yourself: Building a bot that handles tasks in your voice and style
🔒 Governance spotlight: What data is safe to upload to a custom GPT? Understanding the difference between ChatGPT Team/Enterprise vs personal accounts. Data handling across platforms (ChatGPT, Claude, Copilot). How to build useful bots without exposing sensitive or proprietary information. Building guardrails into your bot's system instructions (e.g., "never share client names", "always caveat financial advice").

Live build: Creating a custom GPT from scratch
Also demonstrate: Claude Projects as an alternative approach (for the 34% already using Claude)
Participant challenge: Each person identifies the role or task their bot will serve
Knowledge file preparation: Selecting and uploading relevant documents
Testing protocols: How to validate your bot works as intended
Governance check: Before uploading, ask "would I be comfortable if this document was public?" — if not, it shouldn't go into a consumer AI tool

Build your first custom GPT (this will be part of your final presentation project)
Choose a specific use case: personal assistant, meeting prep bot, content creator, research assistant, etc.
Upload at least 3 relevant knowledge files
Write clear system instructions (using ROSE-T principles)
Test your bot with 10 different queries and document results
Begin refining based on what works and what doesn't

Week3-Homework-Brief
Markdown Cheat Sheet
Clipboard Assistant Template — The free-tier assistant builder (system prompt + knowledge blocks + conversation starters + workflow)
System Instructions Writing Guide — ROSE-T connection, six essential sections, worked example, common mistakes
Knowledge Base Preparation Guide — What to include/avoid, formatting for both tracks, preparation checklist
Custom Assistant Testing Log — Landscape 10-query test protocol with worked example + reflection

Identify high-impact AI use cases within organisations
Map workflows and processes suitable for AI augmentation
Understand the difference between task automation and process transformation
Apply ethical frameworks to proposed use cases
Calculate and present return on investment (ROI)

Use case identification frameworks
Workflow mapping: Understanding current state before introducing AI
Low-hanging fruit vs. transformational opportunities
The AI readiness checklist: When is a process ready for AI?
Common enterprise use cases: Customer service, data analysis, content creation, meeting management, reporting
Red flags: When NOT to use AI
Governance and guardrails: Policy development, data privacy, quality control
Measuring success: Time saved, quality improvements, employee satisfaction

Workflow mapping exercise: Chart a process from your organisation
Identify 3 points where AI could add value
Risk assessment: What could go wrong?
ROI calculation: Estimating time and cost savings

Complete a workflow map for your custom GPT use case
Calculate the time savings your bot could generate (show your working)
Identify any ethical or governance concerns with your proposed use case
Prepare a 2-3 minute pitch: What problem does your bot solve?
Continue refining your custom GPT based on testing

Understand the difference between AI and automation
Recognise when to connect multiple AI agents
Design simple multi-agent workflows
Understand no-code automation platforms (Make.com, Zapier and options including coPilot automations and claude code etc)
Implement quality control and human-in-the-loop checkpoints

What is agentic AI? Understanding autonomous task execution
Automation fundamentals: Triggers, actions, and logic
Popular platforms: Make.com, Zapier, and native AI integrations
Designing multi-step workflows: Email triage → Summary → Calendar → Response
Agent personas: Specialised bots working together
Quality control: When and how to implement human oversight
Example workflows:
- Content creation pipeline: Research → Draft → Edit → Schedule
- Customer enquiry handling: Classify → Route → Respond
- Meeting follow-up: Transcribe → Summarise → Action items → Task assignment

Live demonstration: Building a simple two-agent workflow
Conceptual design: Participants sketch a multi-agent system for their use case
Introduction to API integrations (conceptual—no coding required)

Design a multi-agent workflow on paper (even if you can’t build it yet)
Identify where your custom GPT could “hand off” to another agent
Document the quality control points in your workflow
Finalise your custom GPT for next week’s presentations
Prepare presentation slides (5-7 slides maximum)