Copilot AI World
Go from AI-curious to AI-generalist across every Microsoft 365 app — Word, Excel, PowerPoint, Outlook and Teams — with real labs and real-world outcomes.
Understand & Trust AI
Demystify how AI thinks, master prompting frameworks, and place Copilot in the wider model landscape.
Work Smarter in M365
Apply Copilot across Excel, Power BI, PowerPoint, Word, Teams and SharePoint to compress hours into minutes.
Analyze & Decide
Turn raw data and lengthy PDFs, audio and video into boardroom-ready insights.
Automate & Build
Automate repetitive workflows end to end and build custom agents with Copilot Studio.
Demystifying AI: Building Your AI Foundation
Discover the what, why and how of AI to build a rock-solid foundation and navigate the AI landscape with confidence.
Introduction to AI
- What AI is and how it “thinks”
- AI vs Automation vs Generative AI
- Establishing trust in AI output
The AI Glossary
- ML, DL, LLMs & Agentic AI
- Neural networks explained
- CPU vs GPU vs TPU
- Tokens & the context window
Ethics & Reality
- Hallucinations — why & how to spot
- Public vs Enterprise data mode
- Copyright & IP basics
- The future of AI
Explain core AI concepts in plain business language
Understand why AI makes mistakes and how to avoid hallucinations
Manage the context window for accurate answers
Use AI responsibly with the right data mode
Recognise when an AI answer needs a second check before you act on it.
Keep sensitive information private by choosing the right data mode.
The AI Detective
Spot a hallucination and understand why it happens, estimate token usage for a sample paragraph, and learn best-practice habits for any LLM.
The Art of the Prompt: Engineering Success
Move beyond basic chatting. Learn the frameworks and techniques to command AI like a pro.
Setting the Stage
- Set up custom instructions for Copilot
- Give the AI its role, context and tone up front
- Reusable prompt templates for daily tasks
Core Techniques
- Zero-shot & few-shot prompting
- Chain-of-thought reasoning
- Instruction & reverse prompting
“Perfect Prompt” Frameworks
- Role + Task + Format
- Context + Task + Constraint + Format
- COSTAR: Context, Objective, Steps, Tools, Actions, Reflections
Apply zero-shot, few-shot and chain-of-thought on demand
Build prompts using Role+Task+Format and COSTAR
Set up reusable custom instructions for consistency
Diagnose and rebuild a weak prompt methodically
Make the AI ask clarifying questions before it answers, so output fits your real intent.
Pick the right prompting style for each scenario instead of guessing.
The Prompt Makeover Challenge
Take a vague prompt and evolve it through three frameworks, comparing the output at each stage to see the power of context and constraint.
The Global AI Landscape: Where Copilot Fits
Explore the wider world of AI models to understand where Copilot sits and how it compares.
Right Tool, Right Job
- Matching the right model to each task
- How Copilot compares with other tools
- Which tool suits which purpose
Models inside Copilot
- GPT-5.5 in Copilot Chat
- Claude Opus 4.8 as a model choice
- Reasoning vs fast-response models
The Wider Field
- ChatGPT, Gemini, Claude, Perplexity
- Open-source: Llama, DeepSeek, Qwen
- Picking models by cost & privacy
Understand where Copilot stands against other tools
Choose between fast and reasoning models in Copilot
Decide which AI tool fits a given task
Explain why Copilot suits Microsoft 365 work
Quickly decide which AI tool to open before starting a task.
Switch Copilot to a reasoning model for a hard, multi-step problem.
The Model Showdown
Run one complex reasoning prompt across Copilot’s model options and document the strengths of each for your own work.
The Data Whisperer: Excel & Power BI
Transform raw numbers into actionable insights in seconds with AI-driven analysis, visualisation and forecasting.
Foundations
- Manual analysis using Copilot Chat
- Uploading vs identifying data
- The EDA process
- Predictive vs prescriptive
Copilot in Excel
- Core functions & the Analysis agent
- Plan mode: preview changes first
- Clean, forecast & analyse
- Automated dashboard creation
Advanced Integrations
- Agent mode for multi-step edits
- Python & R in Excel and Power BI
- Best-fit visualisations & dashboards
Clean and prep messy datasets in minutes
Run the full EDA process without manual formulas
Generate predictive insights using Python in Excel
Build automated, reusable dashboards
Turn a messy spreadsheet into a clean, refreshable dashboard in minutes.
Forecast next quarter’s numbers with an AI-generated script — no coding needed.
From Raw Data to Boardroom Dashboard
Use Copilot to clean a messy dataset, generate analytical reports, run a predictive Python script, and build an automated dashboard.
The Master Storyteller: PowerPoint & Designer
Say goodbye to boring slides. Use AI to generate, design and animate stunning presentations and marketing materials.
Copilot in PowerPoint
- Create decks from a prompt or document
- PowerPoint agent: focus, length & theme
- Rewrite, condense & restyle slides
- Brand-aligned images inline
Next-Gen Presentations
- Animated, “Gamma-level” decks
- Document & video → PPT
- Auto speaker notes & presenter coach
Multimedia & Designer
- Designer: brochures, posters, infographics
- Text ↔ video & media from prompts
- Apply themes & templates automatically
Generate a full pitch deck from a single prompt
Convert documents and videos into polished slides
Produce brochures, posters and infographics in Designer
Auto-create speaker notes and animated decks
Produce an investor-ready pitch deck in one sitting instead of over days.
Spin one short brief into a brochure, posters and an infographic in a single session.
The 10-Minute Pitch Deck
Use Designer to generate a brand image, then Copilot in PowerPoint to draft a 5-slide deck from a prompt, restyle the layout and auto-generate speaker notes.
The Document Alchemist: Word & Copilot Notebooks
Synthesise massive amounts of information — turn lengthy documents, audio and video into digestible insights.
Features in Word
- Summarise & analyse documents
- Audio summary in Word
- Transcribe audio or video
- Document → webpage
Copilot Notebooks
- Multi-format: PDF, audio, video, PPT
- Excel & infographics from a notebook
- Audio & video summaries
- Mind maps, reports & study guides
Deep Research
- Deep research & document intelligence
- Multimodal capture on mobile
- Model council & critique
Summarise long PDFs, audio and video into key points
Build study guides and mind maps from mixed sources
Transcribe and analyse meetings and recordings
Turn raw research into a finished Word report
Condense a long document bundle into a one-page brief with the key points.
Turn recorded sessions into searchable study guides and mind maps.
The Deep-Dive Research Synthesizer
Load a lengthy PDF and a video transcript into Copilot Notebooks, generate a study guide and mind map, then transform the findings into a polished Word document.
The Collaborative Architect: Teams, OneDrive & Planner
Supercharge team productivity by letting AI handle project management, files and meeting facilitation.
File & Task Management
- Copilot with OneDrive
- Manage plans & projects in Planner
- Generate Forms with AI
- Build sites & pages by natural language
Microsoft Teams Excellence
- Facilitator agent in Teams
- AI summaries & meeting notes
- Call delegation: Copilot answers calls
- Real-time translation & action items
Capture meeting summaries and action items automatically
Generate a project plan straight from a kickoff call
Organise files and tasks across OneDrive and Planner
Create feedback Forms with AI in seconds
Walk out of a meeting with action items, owners and a draft plan already in place.
Auto-capture meeting notes so no decision is ever lost.
The Ultimate Project Kickoff
Use Teams Copilot to summarise a mock meeting transcript, extract action items, generate a project plan, and create a feedback Form for the team.
The Automation Engine: Building with AI Agents
Become an AI architect. Move beyond built-in tools to creating your own custom agents and workflows.
Pre-Built Agents
- Researcher & Analyst agents
- Prompt coach
- Office agents (Word, Excel, PPT)
- Workflows agent
Custom Automation
- Build agents in Copilot Studio
- Create custom chatbots
- Federated MCP connectors to live data
- Agent 365 governance & A2A
Use pre-built Microsoft agents for research and analysis
Design and deploy a custom chatbot
Automate repetitive workflows end to end
Plan an agent strategy for a team
Deploy a custom agent that reads documents and shortlists results automatically.
Build a workflow agent that routes approvals and updates files without manual chasing.
Build Your First AI Colleague
Design a simple custom chatbot or workflow agent in Copilot Studio — e.g. an FAQ bot or an approvals agent for a repetitive task.
Bring this programme to your team
Each track runs as an instructor-led masterclass with live labs, tailored to your organisation’s tools and data. Talk to us about scheduling and customisation.
Gemini AI World
Master Google’s Gemini across Workspace — Docs, Sheets, Slides, Gmail and Meet — plus Gems, NotebookLM, Deep Research and Veo, with hands-on labs and real outcomes.
Understand & Trust AI
Demystify how AI thinks, master prompting and see where Gemini fits in the model landscape.
Work Smarter in Workspace
Apply Gemini across Gmail, Docs, Sheets, Slides and Meet to compress hours into minutes.
Analyse & Decide
Turn raw data, long PDFs and recordings into insights with NotebookLM and Deep Research.
Automate & Build
Build custom Gems and automate Workspace with agents like Project Mariner and AppSheet.
Demystifying AI: Building Your AI Foundation
Discover the what, why and how of AI to build a rock-solid foundation and navigate the AI landscape with confidence.
Introduction to AI
- What AI is and how it “thinks”
- AI vs Automation vs Generative AI
- Establishing trust in AI output
The AI Glossary
- ML, DL, LLMs & Agentic AI
- Neural networks explained
- CPU vs GPU vs TPU
- Tokens & the context window
Ethics & Reality
- Hallucinations — why & how to spot
- Public vs Workspace data protection
- Copyright & IP basics
- The future of AI
Explain core AI concepts in plain business language
Understand why AI makes mistakes and how to avoid hallucinations
Manage the context window for accurate answers
Use AI responsibly with the right data mode
Recognise when an AI answer needs a second check before you act on it.
Keep sensitive information private by choosing the right data mode.
The AI Detective
Spot a hallucination and understand why it happens, estimate token usage for a sample paragraph, and learn best-practice habits for any LLM.
The Art of the Prompt: Engineering Success
Move beyond basic chatting. Learn the frameworks and techniques to command AI like a pro.
Setting the Stage
- Set up saved info & custom instructions
- Give Gemini its role, context and tone
- Reusable prompt templates for daily tasks
Core Techniques
- Zero-shot & few-shot prompting
- Chain-of-thought reasoning
- Instruction & reverse prompting
“Perfect Prompt” Frameworks
- Role + Task + Format
- Context + Task + Constraint + Format
- COSTAR: Context, Objective, Steps, Tools, Actions, Reflections
Apply zero-shot, few-shot and chain-of-thought on demand
Build prompts using Role+Task+Format and COSTAR
Set up reusable custom instructions for consistency
Diagnose and rebuild a weak prompt methodically
Make the AI ask clarifying questions before it answers, so output fits your real intent.
Pick the right prompting style for each scenario instead of guessing.
The Prompt Makeover Challenge
Take a vague prompt and evolve it through three frameworks, comparing the output at each stage to see the power of context and constraint.
The Global AI Landscape: Where Gemini Fits
Explore the wider world of AI models to understand where Gemini sits and how it compares.
Right Tool, Right Job
- Matching the right model to each task
- How Gemini compares with other tools
- Which tool suits which purpose
The Gemini Family
- Gemini 3.1 Pro & Flash
- 1M-token context window
- Thinking / reasoning modes
- AI Plus, Pro & Ultra tiers
The Wider Field
- ChatGPT, Copilot, Claude, Perplexity
- Open-source: Llama, DeepSeek, Gemma
- Picking models by cost & privacy
Understand where Gemini stands against other tools
Choose between Flash and Pro for the task
Decide which AI tool fits a given task
Explain why Gemini suits Google Workspace work
Quickly decide which AI tool to open before starting a task.
Switch to a thinking model for a hard, multi-step problem.
The Model Showdown
Run one complex reasoning prompt across Gemini’s models and document the strengths of each for your own work.
The Data Whisperer: Gemini in Sheets
Turn raw numbers into actionable insights with Gemini’s side panel in Google Sheets.
Foundations
- Manual analysis in the Gemini side panel
- Uploading vs referencing data
- The EDA process
- Predictive vs prescriptive
Gemini in Sheets
- Formulas in natural language
- Clean, categorise & forecast
- Charts & pivots from prompts
- “Help me organise” tables
Advanced
- Connected Sheets & BigQuery basics
- Deep Research Max for data (+MCP)
- Native visualisations & dashboards
Clean and prep messy datasets in minutes
Run the full EDA process without manual formulas
Generate predictive insights from your data
Build automated, shareable dashboards
Turn a messy spreadsheet into a clean, refreshable dashboard in minutes.
Forecast next quarter’s numbers from your Sheets data — no coding needed.
From Raw Data to Boardroom Dashboard
Use Gemini in Sheets to clean a messy dataset, generate formulas and reports, forecast, and build an automated dashboard.
The Master Storyteller: Slides, Veo & Imagen
Generate, design and animate stunning presentations, images and video.
Gemini in Slides
- Generate slides & speaker notes
- “Help me visualise” images inline
- Restyle & summarise decks
Image & Video
- Imagen / Nano Banana Pro images
- Veo 3.1 text- & image-to-video
- Google Flow for AI filmmaking
Multimedia
- Audio Overviews from NotebookLM
- Brand-aligned visuals from prompts
- Whisk & Ingredients-to-Video
Generate a full deck from a single prompt
Convert documents into polished slides
Produce on-brand images and short videos
Auto-create speaker notes and visuals
Produce a client-ready deck in one sitting instead of over days.
Spin one short brief into images and a short promo video in a single session.
The 10-Minute Pitch Deck
Generate a deck in Slides, create a brand image with Imagen and a short clip with Veo, then auto-generate speaker notes.
The Document Alchemist: NotebookLM & Deep Research
Synthesise massive information into digestible insights with NotebookLM and Deep Research.
NotebookLM
- Upload hundreds of sources
- Audio Overviews (podcast-style)
- Mind maps, study guides & briefs
- Cited answers grounded in sources
Deep Research
- Autonomous multi-source research
- Deep Research Max: charts + MCP
- Guide the research plan
- Reports with citations
Multi-format
- PDF, video, audio & Docs
- Gemini in Chrome page summaries
- Guided Learning
Summarise long PDFs, audio and video into key points
Build study guides and mind maps from mixed sources
Run deep, cited research autonomously
Turn raw research into a finished report
Condense a long document bundle into a one-page brief with the key points.
Turn recorded sessions into searchable study guides and mind maps.
The Deep-Dive Research Synthesizer
Load sources into NotebookLM, generate a study guide, mind map and Audio Overview, then run Deep Research for a cited report.
The Collaborative Architect: Gmail, Meet & Workspace
Let Gemini handle inbox, meetings and documents across Workspace.
Gmail & Docs
- Summarise & draft emails in Gmail
- “Help me write” & refine in Docs
- Smart replies & tone control
Google Meet
- “Take notes for me” & summaries
- Real-time translated captions
- Action items & recaps
Files & Forms
- Find & summarise files in Drive
- Generate Google Forms
- Chat across your Workspace content
Capture meeting summaries and action items automatically
Draft and triage email in seconds
Find and summarise files across Drive
Create feedback Forms with AI in seconds
Walk out of a Meet call with action items and a recap already written.
Clear your inbox faster with AI summaries and drafted replies.
The Ultimate Project Kickoff
Use Meet to summarise a mock meeting and extract actions, then draft the follow-up email in Gmail and a plan in Docs.
The Automation Engine: Gems, Mariner & AppSheet
Build custom Gems and automate Workspace with Google’s agents.
Custom Gems
- Create reusable expert Gems
- Knowledge & instructions per Gem
- Share Gems with your team
Agents & Automation
- Project Mariner: agentic browsing
- Gemini in AppSheet: apps from a description
- Workspace Studio automation flows
Build a reusable Gem for a repeatable task
Design a no-code app from a description
Automate a multi-step Workspace workflow
Plan an agent strategy for a team
Build a custom Gem that drafts proposals in your house style.
Spin up a no-code app for an internal process without IT.
Build Your First AI Colleague
Create a custom Gem with its own knowledge and instructions, or build a simple AppSheet app for a team process.
Bring this programme to your team
Each track runs as an instructor-led masterclass with live labs, tailored to your organisation’s tools and data. Talk to us about scheduling and customisation.
ChatGPT AI World
Master OpenAI’s ChatGPT end to end — GPT-5.5, Custom GPTs, Canvas, image generation, voice, Deep Research and Agent mode — with real labs and outcomes.
Understand & Trust AI
Demystify how AI thinks, master prompting and place ChatGPT in the model landscape.
Work Smarter Daily
Use Projects, Tasks, memory, Canvas and voice to compress everyday work.
Analyse & Decide
Analyse data and files and run Deep Research into clear, sourced reports.
Automate & Build
Build Custom GPTs, schedule Tasks, and run Agent mode and Codex.
Demystifying AI: Building Your AI Foundation
Discover the what, why and how of AI to build a rock-solid foundation and navigate the AI landscape with confidence.
Introduction to AI
- What AI is and how it “thinks”
- AI vs Automation vs Generative AI
- Establishing trust in AI output
The AI Glossary
- ML, DL, LLMs & Agentic AI
- Neural networks explained
- CPU vs GPU vs TPU
- Tokens & the context window
Ethics & Reality
- Hallucinations — why & how to spot
- Personal vs business data & training opt-out
- Copyright & IP basics
- The future of AI
Explain core AI concepts in plain business language
Understand why AI makes mistakes and how to avoid hallucinations
Manage the context window for accurate answers
Use AI responsibly with the right data mode
Recognise when an AI answer needs a second check before you act on it.
Keep sensitive information private by choosing the right data mode.
The AI Detective
Spot a hallucination and understand why it happens, estimate token usage for a sample paragraph, and learn best-practice habits for any LLM.
The Art of the Prompt: Engineering Success
Move beyond basic chatting. Learn the frameworks and techniques to command AI like a pro.
Setting the Stage
- Set up custom instructions & memory
- Give ChatGPT its role, context and tone
- Reusable prompt templates for daily tasks
Core Techniques
- Zero-shot & few-shot prompting
- Chain-of-thought reasoning
- Instruction & reverse prompting
“Perfect Prompt” Frameworks
- Role + Task + Format
- Context + Task + Constraint + Format
- COSTAR: Context, Objective, Steps, Tools, Actions, Reflections
Apply zero-shot, few-shot and chain-of-thought on demand
Build prompts using Role+Task+Format and COSTAR
Set up reusable custom instructions for consistency
Diagnose and rebuild a weak prompt methodically
Make the AI ask clarifying questions before it answers, so output fits your real intent.
Pick the right prompting style for each scenario instead of guessing.
The Prompt Makeover Challenge
Take a vague prompt and evolve it through three frameworks, comparing the output at each stage to see the power of context and constraint.
The Global AI Landscape: Where ChatGPT Fits
Explore the wider world of AI models to understand where ChatGPT sits and how it compares.
Right Tool, Right Job
- Matching the right model to each task
- How ChatGPT compares with other tools
- Which tool suits which purpose
The GPT Family
- GPT-5.5 Instant & Thinking
- Auto model routing
- Reasoning vs fast models
- Free, Go, Plus & Pro tiers
The Wider Field
- Gemini, Copilot, Claude, Perplexity
- Open-source: Llama, DeepSeek, Qwen
- Picking models by cost & privacy
Understand where ChatGPT stands against other tools
Choose between Instant and Thinking models
Decide which AI tool fits a given task
Use Auto routing sensibly
Quickly decide which AI tool to open before starting a task.
Switch to a Thinking model for a hard, multi-step problem.
The Model Showdown
Run one complex reasoning prompt across ChatGPT’s models and document the strengths of each for your own work.
The Data Whisperer: Data Analysis & Code Interpreter
Transform raw numbers into insights with ChatGPT’s built-in Python data analysis.
Foundations
- Manual analysis in chat
- Uploading vs identifying data
- The EDA process
- Predictive vs prescriptive
Data Analysis
- Upload CSV / XLSX to analyse
- Python sandbox (Code Interpreter)
- Charts, pivots & cleaning
- Statistical summaries
Advanced
- Predictive scripts & forecasts
- Connect data via connectors
- Export polished reports
Clean and prep messy datasets in minutes
Run the full EDA process without manual formulas
Generate predictive insights using Python
Export analysis as a clear report
Turn a messy spreadsheet into clean charts and a summary in minutes.
Forecast next quarter’s numbers with an AI-generated Python script.
From Raw Data to Boardroom Dashboard
Upload a messy dataset, use Code Interpreter to clean it, generate charts and a forecast, then export a polished report.
The Master Storyteller: Images, Canvas & Voice
Generate visuals, draft polished content and work hands-free with voice.
Image Generation
- GPT Image native generation
- Editing & strong prompt adherence
- Flyers, social & brand visuals
Canvas & Writing
- Writing & code blocks (Canvas)
- Side-by-side editing & iteration
- Rewrite, condense & restyle
Voice & Browse
- Advanced Voice Mode & translation
- Slide outlines from a brief
- Atlas browser assistance
Generate and edit on-brand images from prompts
Draft and refine long content in Canvas
Work hands-free with Advanced Voice
Produce slide outlines from a brief
Create a set of on-brand social graphics from one prompt.
Draft and polish a long article side-by-side in Canvas.
The Creative Workshop
Generate a brand image, draft a one-page document in Canvas, and refine it hands-free using Advanced Voice Mode.
The Document Alchemist: Deep Research & Files
Synthesise large amounts of information into clear, sourced output.
Deep Research
- Autonomous web research with citations
- Multi-source synthesis
- Structured reports
- Run limits by tier
Document Intelligence
- Analyse PDF, DOCX, XLSX, images
- Summarise long documents
- Q&A across uploads
Knowledge & Memory
- Projects for grouped context
- Persistent memory
- Cited, checkable answers
Run deep, cited research autonomously
Summarise and query long documents
Group context and files in Projects
Turn research into a structured report
Produce a cited research brief on a topic in one run.
Ask questions across a stack of PDFs and get sourced answers.
The Deep-Dive Research Synthesizer
Upload a long PDF, run Deep Research on a related question, and combine both into a cited, structured brief.
The Productivity Architect: Projects, Tasks & GPT Store
Organise your work and let scheduled automations run for you.
Projects & Tasks
- Group chats & files into Projects
- Scheduled Tasks & reminders
- Per-project instructions & memory
Team & Sharing
- Custom GPTs for repeatable work
- GPT Store: find & share assistants
- Business workspace & data privacy
Organise work into Projects with shared context
Schedule recurring Tasks
Reuse Custom GPTs for repeatable work
Keep business data private in workspaces
Get a daily briefing delivered automatically by a scheduled Task.
Keep every client’s context in its own Project.
The Workflow Kickoff
Create a Project with custom instructions and files, then set up a scheduled Task that delivers a recurring summary.
The Automation Engine: Agent Mode, Codex & GPTs
Move beyond chatting to building assistants and agents that take action.
Custom GPTs
- Build task-specific GPTs
- Knowledge files & tools
- Publish to the GPT Store
Agents & Automation
- ChatGPT Agent mode (computer use)
- Codex for agentic coding
- Connectors & end-to-end workflows
Build and publish a Custom GPT
Use Agent mode to complete a task end to end
Apply Codex to a coding task
Plan an agent strategy for a team
Build a Custom GPT that answers staff FAQs from your documents.
Use Agent mode to research and fill a form without manual steps.
Build Your First AI Colleague
Create a Custom GPT with knowledge files, then run a simple task in Agent mode from start to finish.
Bring this programme to your team
Each track runs as an instructor-led masterclass with live labs, tailored to your organisation’s tools and data. Talk to us about scheduling and customisation.
Claude AI World
Master Anthropic’s Claude end to end — Opus 4.8, Projects, Artifacts, MCP connectors, Claude Code and Cowork — with hands-on labs and real-world outcomes.
Understand & Trust AI
Demystify how AI thinks, master prompting and place Claude in the model landscape.
Work Smarter Daily
Use Projects, memory and Artifacts to compress everyday knowledge work.
Analyse & Decide
Analyse data, Excel and long documents into clear, well-reasoned output.
Automate & Build
Connect tools with MCP and automate work with Cowork and Claude Code.
Demystifying AI: Building Your AI Foundation
Discover the what, why and how of AI to build a rock-solid foundation and navigate the AI landscape with confidence.
Introduction to AI
- What AI is and how it “thinks”
- AI vs Automation vs Generative AI
- Establishing trust in AI output
The AI Glossary
- ML, DL, LLMs & Agentic AI
- Neural networks explained
- CPU vs GPU vs TPU
- Extended thinking & reasoning
- Tokens & the context window
Ethics & Reality
- Hallucinations — why & how to spot
- Consumer vs commercial data & privacy
- Copyright & IP basics
- The future of AI
Explain core AI concepts in plain business language
Understand why AI makes mistakes and how to avoid hallucinations
Manage the context window for accurate answers
Use AI responsibly with the right data mode
Recognise when an AI answer needs a second check before you act on it.
Keep sensitive information private by choosing the right data mode.
The AI Detective
Spot a hallucination and understand why it happens, estimate token usage for a sample paragraph, and learn best-practice habits for any LLM.
The Art of the Prompt: Engineering Success
Move beyond basic chatting. Learn the frameworks and techniques to command AI like a pro.
Setting the Stage
- Set Project instructions, styles & memory
- Give Claude its role, context and tone
- Structure prompts with clear sections (XML-style)
Core Techniques
- Zero-shot & few-shot prompting
- Chain-of-thought reasoning
- Instruction & reverse prompting
“Perfect Prompt” Frameworks
- Role + Task + Format
- Context + Task + Constraint + Format
- COSTAR: Context, Objective, Steps, Tools, Actions, Reflections
Apply zero-shot, few-shot and chain-of-thought on demand
Build prompts using Role+Task+Format and COSTAR
Set up reusable custom instructions for consistency
Diagnose and rebuild a weak prompt methodically
Make the AI ask clarifying questions before it answers, so output fits your real intent.
Pick the right prompting style for each scenario instead of guessing.
The Prompt Makeover Challenge
Take a vague prompt and evolve it through three frameworks, comparing the output at each stage to see the power of context and constraint.
The Global AI Landscape: Where Claude Fits
Explore the wider world of AI models to understand where Claude sits and how it compares.
Right Tool, Right Job
- Matching the right model to each task
- How Claude compares with other tools
- Which tool suits which purpose
The Claude Family
- Opus 4.8, Sonnet 4.6, Haiku 4.5
- 1M-token context (Sonnet)
- Extended thinking & effort control
- Free, Pro & Max tiers
The Wider Field
- ChatGPT, Gemini, Copilot, Perplexity
- Open-source: Llama, DeepSeek, Qwen
- Picking models by cost & privacy
Understand where Claude stands against other tools
Choose between Opus, Sonnet and Haiku
Decide which AI tool fits a given task
Tune reasoning effort for the job
Pick Haiku for speed or Opus for a hard, nuanced problem.
Decide when Claude is the right tool versus an alternative.
The Model Showdown
Run one complex reasoning prompt across Claude’s models and document the strengths of each for your own work.
The Data Whisperer: Analysis Tool & Claude for Excel
Turn raw numbers into insights you can trust, in chat and inside Excel.
Foundations
- Manual analysis in chat
- Uploading vs identifying data
- The EDA process
- Predictive vs prescriptive
The Analysis Tool
- Run code to analyse data
- Charts & summaries from CSV/XLSX
- Clean & reshape data
- Honest, checkable numbers
Claude for Excel
- Edit & build workbooks in-app
- Formulas & multi-step changes
- MCP connectors inside Excel
Clean and prep messy datasets in minutes
Run analysis with verifiable, checkable results
Build and edit workbooks with Claude for Excel
Generate charts and summaries from raw data
Get accurate, checkable numbers from a messy dataset.
Make multi-step edits to a workbook without writing formulas by hand.
From Raw Data to Boardroom Dashboard
Use the Analysis tool to clean a dataset and chart it, then build the summary workbook with Claude for Excel.
The Master Storyteller: Artifacts & Claude Design
Build interactive, on-brand visuals, tools and slides — then iterate by chat.
Artifacts
- Live React, HTML & SVG outputs
- Interactive apps, tools & dashboards
- Persistent storage & AI-powered apps
Design & Slides
- Claude Design canvas
- Claude for PowerPoint decks
- Diagrams, charts & visuals inline
Iterate
- Generate interactive explainers
- Brand-aligned visual components
- Refine everything by chat
Build an interactive tool or app as an Artifact
Create slides with Claude for PowerPoint
Produce diagrams and charts inline
Iterate on a design through conversation
Build a working calculator or dashboard from a description.
Turn a brief into a polished, editable deck.
The Interactive Build
Create an interactive Artifact (e.g. a calculator or mini-dashboard), then generate a short deck with Claude for PowerPoint.
The Document Alchemist: Projects & Deep Research
Synthesise large amounts of information with persistent context and cited research.
Projects
- Persistent workspaces with files
- Project knowledge & instructions
- Search past chats & memory
Research & Documents
- Web search & Deep Research reports
- Analyse PDFs, docs & images
- High-res vision for diagrams & scans
Synthesis
- Summaries, study guides & briefs
- Cited, source-grounded answers
Keep documents and context in a Project
Run cited Deep Research on a topic
Analyse PDFs, scans and diagrams accurately
Turn research into a finished brief
Keep a whole matter’s files in one persistent Project.
Produce a cited research brief grounded in your own sources.
The Deep-Dive Research Synthesizer
Load sources into a Project, run Deep Research on a related question, and produce a cited, structured brief.
The Connected Architect: MCP & Cowork
Connect Claude to the tools you already use and let it work across your files.
MCP Connectors
- Connect Gmail, Drive, Slack, Notion, GitHub
- 200+ MCP servers
- MCP Apps: Canva, Asana, Figma
- Act across your tools
Cowork
- Desktop agent on your folders
- Read, edit & create files autonomously
- Scheduled tasks & sub-agents
Connect Claude to your everyday tools via MCP
Let Claude work across a folder with Cowork
Schedule recurring agent tasks
Coordinate sub-agents on bigger work
Have Claude pull from Drive and Slack to draft a status update.
Point Cowork at a folder to triage and reorganise files.
The Connected Workspace
Connect an MCP tool (e.g. Drive or Slack) and run a Cowork task that reads and updates files in a folder.
The Automation Engine: Claude Code, Skills & Agents
Move beyond chatting to building agents and automations under your control.
Build & Automate
- Claude Code: agentic coding in the terminal
- Skills & custom capabilities
- Computer use in Cowork & Code
Agents at Scale
- Agent Teams & Managed Agents
- Routines & dynamic workflows
- Plan an agent strategy for a team
Use Claude Code for an agentic coding task
Package a repeatable capability as a Skill
Apply computer use to automate steps
Plan an agent strategy for a team
Delegate a multi-file coding change to Claude Code.
Build a reusable Skill your whole team can call.
Build Your First AI Colleague
Create a simple Skill or run an agentic task in Claude Code, then plan how a team could reuse it.
Bring this programme to your team
Each track runs as an instructor-led masterclass with live labs, tailored to your organisation’s tools and data. Talk to us about scheduling and customisation.
Open Source AI World
Run powerful AI on your own terms — Llama, DeepSeek, Qwen, Mistral and Gemma — self-hosted with Ollama and LM Studio for privacy, control and zero per-seat cost.
Understand & Trust AI
Demystify how open models work, master prompting and read benchmarks & licences.
Own Your Stack
Run models on your own hardware for privacy, compliance and predictable cost.
Analyse & Decide
Analyse data and documents locally with private RAG — nothing leaves your machine.
Automate & Build
Fine-tune, build local agents and self-host AI for your whole team.
Demystifying AI: Open Models & Your AI Foundation
Understand how open-weight models work and why self-hosting matters.
Introduction to AI
- What AI is and how it “thinks”
- Open-weight vs closed models
- Why run AI yourself
The AI Glossary
- LLMs, parameters & MoE
- Quantisation (Q4, GGUF)
- Context window & tokens
- GPU, VRAM & CPU inference
Ethics & Licences
- Apache 2.0, MIT & custom licences
- Data privacy & on-prem control
- Hallucinations & how to spot them
Explain open-weight models in plain language
Read a model card, benchmark and licence
Understand quantisation and hardware needs
Decide when self-hosting beats a cloud tool
Keep confidential data fully in-house by running a model offline.
Pick a model whose licence actually allows your commercial use.
The AI Detective
Compare an open and a closed model on the same prompt, read a Hugging Face model card and licence, and estimate the hardware a model needs.
The Art of the Prompt: Engineering Success
Move beyond basic chatting. Learn the frameworks and techniques to command AI like a pro.
Setting the Stage
- System prompts & personas
- Give the model its role, context and tone
- Reusable prompt templates for daily tasks
Core Techniques
- Zero-shot & few-shot prompting
- Chain-of-thought reasoning
- Instruction & reverse prompting
“Perfect Prompt” Frameworks
- Role + Task + Format
- Context + Task + Constraint + Format
- COSTAR: Context, Objective, Steps, Tools, Actions, Reflections
Apply zero-shot, few-shot and chain-of-thought on demand
Build prompts using Role+Task+Format and COSTAR
Set up reusable custom instructions for consistency
Diagnose and rebuild a weak prompt methodically
Make the AI ask clarifying questions before it answers, so output fits your real intent.
Pick the right prompting style for each scenario instead of guessing.
The Prompt Makeover Challenge
Take a vague prompt and evolve it through three frameworks, comparing the output at each stage to see the power of context and constraint.
The Open-Source Landscape: Choosing a Model
Navigate the fast-moving world of open models and pick the right one.
Right Model, Right Job
- Match model to task, licence & hardware
- Dense vs Mixture-of-Experts
- Reading LiveBench / benchmarks
Top Open Models (2026)
- Llama 4 · DeepSeek V4
- Qwen 3.6 · Mistral & Devstral
- Gemma 4 · Phi-4 · Kimi & GLM
Where They Fit
- vs ChatGPT, Gemini, Copilot, Claude
- Cost & privacy trade-offs
- Indian-language & India-region models
Match an open model to task, licence and hardware
Tell dense and MoE models apart
Read benchmarks to compare models
Choose models that fit your budget and privacy needs
Choose a model that runs well on the GPU you actually have.
Avoid licence traps before shipping a commercial app.
The Model Showdown
Pull two open models with Ollama and run the same prompt, documenting speed, quality and licence for each.
The Data Whisperer: Private Data Analysis
Analyse data locally with open models and Python — nothing leaves your machine.
Foundations
- Manual analysis via local chat (Open WebUI)
- Upload vs reference data
- The EDA process
- Predictive vs prescriptive
Local Tooling
- Python + pandas with a local model
- Code generation & cleaning
- Charts & summaries offline
Advanced
- Connect a model to a spreadsheet
- Predictive scripts & forecasts
- Private dashboards
Analyse data without it leaving your machine
Clean and reshape data with a local model
Generate charts and forecasts offline
Build a private dashboard
Analyse sensitive financial or HR data fully offline.
Forecast numbers locally with no cloud dependency.
From Raw Data to Dashboard
Use a local model with Python to clean a dataset, chart it and produce a forecast — all offline.
The Creative Studio: Open Image & Media Models
Generate images and media with open models, fully under your control.
Open Image Models
- Stable Diffusion & Flux
- ComfyUI & Automatic1111
- Flyers, social assets & visuals
Customisation
- LoRAs & fine-tuned styles
- Prompting for consistent brand looks
- Run image gen on local GPU
Multimedia
- Open text-to-speech & voice
- Open transcription (Whisper)
- Local, licence-clean media
Generate images locally with open models
Apply LoRAs for a consistent brand style
Transcribe audio privately with Whisper
Keep all media generation licence-clean
Produce a brand image set on your own GPU, royalty-free.
Transcribe confidential recordings without sending them to the cloud.
The Local Creative Lab
Generate on-brand images in ComfyUI and transcribe a recording with Whisper — entirely on your own machine.
The Document Alchemist: Local RAG & Document Intelligence
Build private retrieval over your own documents — answers grounded in your data.
Local RAG
- Retrieval-Augmented Generation explained
- Embeddings & vector databases
- Chunking & grounded answers
Tools
- AnythingLLM & Open WebUI knowledge
- Chat with PDFs, docs & wikis
- Citations to your own sources
Privacy
- 100% on-prem, offline option
- No data sent to third parties
- Summaries, briefs & study guides
Explain how RAG grounds answers in your data
Build a local knowledge base over your documents
Get cited answers from your own sources
Keep the whole pipeline on-prem
Build a private chatbot over internal documents and policies.
Research a document set offline with citations you can trust.
Build a Private Document Brain
Load a set of documents into a local RAG tool and query them with citations — fully offline.
The Self-Hosting Architect: Running AI for Your Team
Serve open models reliably to a whole team, on your own infrastructure.
Local Tools
- Ollama (CLI) & LM Studio (GUI)
- llama.cpp & quantised builds
- Open WebUI: a ChatGPT-style UI
Production Serving
- vLLM & SGLang for throughput
- GPU tiers: L4, L40S, A100, H100
- Kubernetes for scale
Governance
- Access, logging & cost control
- On-prem vs private cloud
- Predictable per-team economics
Run models locally with Ollama and LM Studio
Serve a model to colleagues via Open WebUI
Choose a serving stack and GPU tier
Govern access and control cost
Give a team a private, ChatGPT-style assistant on your own servers.
Cap AI cost predictably with no per-seat fees.
Stand Up a Team Assistant
Run a model in Ollama and expose it through Open WebUI so colleagues can use it from the browser.
The Automation Engine: Fine-Tuning & Local Agents
Customise models and build agents that run entirely under your control.
Fine-Tuning
- When to fine-tune vs RAG vs prompt
- LoRA / QLoRA basics
- Prepare data & evaluate
Local Agents
- Tool calling & function calling
- OpenCode + Ollama coding agent
- Agent frameworks & workflows
Decide between fine-tuning, RAG and prompting
Run a basic LoRA fine-tune
Build a tool-calling local agent
Plan a self-hosted agent strategy
Fine-tune a small model to write in your house style.
Run a local coding agent that automates a repetitive task offline.
Build Your First Local Agent
Set up a tool-calling agent or an OpenCode + Ollama coding assistant that runs on your own hardware.
Bring this programme to your team
Each track runs as an instructor-led masterclass with live labs, tailored to your organisation’s tools and data. Talk to us about scheduling and customisation.