Claude Code for Non-Developers Training Course

Phase 1 — Meet Claude Code — 55 minutes

• What Claude is, and the distinctions that set Claude Code apart from standard chat
• The Claude product family: claude.ai, Claude Desktop, Claude Code (CLI), and their relationships
• Interface overview: navigating the Claude app, initiating a coding session, and understanding the workspace
• How Claude Code operates: the describe → plan → act → review cycle
• Understanding permissions: the rationale behind Claude asking before creating files or executing code
• Your first build: asking Claude to create a simple styled webpage from a one-sentence description
• Refining results: “make the header larger,” “change the colour scheme,” “add a navigation bar”
• Guided exercise: Participants open the Claude app, start a Claude Code session, and build a personalised “About Me” webpage by describing their requirements in plain English. They practice refining their output through follow-up instructions.

Objective: Everyone is comfortable with the interface and has overcome the initial interaction barrier.

Break — 10 minutes

Phase 2 — Building Real Things with Plain English — 70 minutes

This is the core of the morning. Participants complete four progressively complex tasks using only natural language prompts.

• Task 1 — Interactive dashboard: Ask Claude Code to build a styled dashboard displaying sample data with charts, statistics, and a clean layout. Practice providing design direction: “use a dark theme,” “add a sidebar,” “make it responsive.”
• Task 2 — Data analysis: Provide Claude with a sample CSV file and ask it to summarise the data, identify trends, find the highest and lowest values, and generate a visual chart. This demonstrates how Claude writes and executes code on your behalf.
• Task 3 — Document generator: Ask Claude to read a data file and produce a formatted report — a sales summary, a project status update, or a meeting recap. This shows how Claude can transform raw data into polished deliverables.
• Task 4 — Automation tool: Ask Claude to build a simple utility — a unit converter, a quiz app, or a budget calculator. This introduces the concept that Claude can build interactive tools, not just static pages.

After each task, the instructor highlights what Claude did behind the scenes: which files it created, what code it wrote, and how to interpret the output. Participants document their most effective prompts in a shared Prompt Playbook.

Break — 10 minutes

Phase 3 — Working Smarter with Claude Code — 50 minutes

• The art of effective prompting: specific versus vague instructions
• Live demonstration: side-by-side comparison of weak and strong prompts on the same task
• Iterating and refining: asking Claude to explain its choices, undo changes, or try a different approach
• Working with uploaded files: “read this document and summarise it,” “convert this spreadsheet into a chart”
• Multi-step workflows: chaining requests to build complex outputs (“first analyse this data, then build a dashboard from the results”)
• Understanding cost and usage: how tokens, context windows, and subscription tiers function
• When to use Claude Code versus regular Claude chat
• Guided exercise: Participants take one of their Phase 2 projects and extend it with two new features using a multi-step prompt chain. They then compare their before-and-after prompts to identify what made the difference.

Objective: Level up from “it works” to “I can get great results consistently.”

Break — 10 minutes

Phase 4 — Your Claude Workflows: Live Build Session — 60 minutes

This phase shifts the energy in the room. Instead of individual practice, the group builds together. The instructor guides the process, but participants direct the outcome — naming real problems from their own roles, suggesting prompt ideas, and debating trade-offs. The aim is to learn prompt judgment by observing a skilled practitioner navigate uncertainty in real time.

Three workflow archetypes structure the session:

• Transform — take input X, produce output Y (meeting notes → action items; raw data → summary email; customer feedback → themed report)
• Draft — generate a first version of something you’d normally write from scratch (proposals, emails, job descriptions, social posts)
• Analyse — interrogate a document or dataset you don’t have time to read carefully (a 40-page report, a spreadsheet of survey responses, a contract)

Setup and framing (10 min): The instructor introduces the three archetypes and explains how the session works. Participants submit real workflow problems from their jobs via a shared document or chat.

Live build #1 — Transform workflow (20 min): The instructor selects one submitted problem and builds it live, with the room calling out prompt ideas, pushbacks, and refinements. The instructor narrates every choice. The session ends with a working prompt template that the participant whose problem it was gets to keep.

Live build #2 — Draft or Analyse workflow (20 min): Same format, different archetype, different participant’s problem.

Reflection & share-back (10 min): Participants take a moment to note down one prompting move that surprised them, one thing they’d do differently, and one pattern they’re taking home. Quick group share — 3-4 voices, not everyone. The instructor connects observations to the broader Prompt Playbook.

Phase 5 — Connecting Claude to Your Tools with MCP — 50 minutes

• What is MCP (Model Context Protocol)? The universal plug system for AI tools
• Why MCP matters: turning Claude from a chat assistant into a connected workflow hub
• The Connectors Directory: browsing and adding integrations directly from the Claude app
• Desktop Extensions: one-click installs for Claude Desktop (no configuration files needed)

Live demonstration: The instructor connects Claude to two services through the Connectors UI and demonstrates cross-tool workflows:

1. “Check my Google Calendar for tomorrow’s meetings and draft a prep email for each one”
2. “Read the latest updates from our project board and write a status summary”
3. “Pull data from this connected service and build a local report from it”

Guided exercise: Participants connect Claude to at least one service. Options are provided for different comfort levels:

• Option A: Connect a pre-built connector from the directory (e.g., Gmail, Google Drive, or a demo service) — click, authenticate, and go
• Option B: Add a custom connector by pasting an MCP server URL (instructor provides a test URL)
• Option C: Install a Desktop Extension from the marketplace (Claude Desktop users)

Participants then give Claude a task that uses the connected service — e.g., “Read my recent emails about project updates and create a summary document.”

Key concepts covered:

• How connectors work: OAuth authentication, permissions, and the level of access being granted
• Managing tool access: enabling, disabling, and controlling which connectors Claude can use per conversation
• Security awareness: connecting only to trusted services and reviewing tool permissions
• The MCP ecosystem: where to find new connectors, extensions, and community-built servers

Objective: Participants view Claude as a connective layer between all the services they already use, not just a coding tool.

Break — 10 minutes

Phase 6 — Capstone & Next Steps — 65 minutes

Capstone mini-project (45 min): Each participant chooses one scenario and builds it with Claude:

1. A polished landing page or portfolio site for their team, project, or personal brand
2. A data analysis pipeline: upload a file, have Claude analyse it, and produce a visual report
3. An interactive tool that solves a real problem from their workflow (calculator, tracker, converter, quiz)
4. A connected workflow: pull data from a connected service, transform it, and produce a deliverable (e.g., “read my calendar for next week and build a visual schedule”)

The instructor circulates, helps refine prompts, and showcases standout examples to the group.

Showcase and wrap-up (20 min):

• 6-8 participants share what they built (2-3 min each)
• Where to go from here: Claude Code CLI for terminal users, VS Code extension for developers, Cowork for knowledge workers
• The MCP ecosystem: finding and evaluating new connectors, extensions, and community servers
• Plans: Free vs. Pro vs. Max — what each unlocks and which fits which use case
• Best practices recap: the Prompt Playbook patterns that worked best during the session
• Recommended resources: official documentation, community channels, Anthropic’s prompt engineering guide
• Participants receive a reference card with key prompting patterns, connector setup steps, and a curated list of useful MCP integrations