Fundamentals instead of a black box
Layer 0 covers LLMs, hallucinations, knowledge cutoff and context window — participants know what they compare when testing models.
Claude + MCP · Introduction to AI
Claude + MCP: introduction to working with AI
A two-day programme for teams adopting Claude — from LLM fundamentals and privacy settings to surfaces (Chat, Cowork, Design, Code), Projects, Connectors/MCP, Skills and responsible use in the organisation.
Who it is for
Designers, team leads, operations specialists and non-technical staff who will use Claude daily — plus organisations planning Team/Enterprise rollout with admin, roles and privacy policy.
Benefits
Enterprise and individual in one programme
From Pro/Max to Team/Enterprise: licensing, privacy, retention, ZDR, admin roles — without mixing personal and company settings.
From chat to ecosystem
Projects, Connectors, MCP, Skills, Memory and Artifacts — not a feature list, but a coherent workflow for design, ops and dev roles.
What you get
- You understand how an LLM answer is produced, where models help vs. hallucinate — and test Opus, Sonnet and Haiku on the same tasks
- You configure accounts, licensing (Pro/Max vs Team/Enterprise), privacy, data retention and ZDR — individually and for teams
- You pick the right Claude surface (decision map) and connect apps via Connectors and MCP
- You build durable context: Projects, Skills, Memory, Artifacts — with rules for client data and output verification
Outcomes
- Shared language: when AI helps, when to verify, when it should not decide
- A ready map: which Claude surface (Chat, Cowork, Design, Code) for which task type
- First app connections (Gmail, Drive, Calendar) and a conscious choice of Skills / Projects / MCP
- Privacy, cost and sensitive-data checklists — before the team "dumps everything into chat"
The programme is layered: **Layer 0 (LLM fundamentals)** is the root — without it, model testing and tool choice float in a vacuum. Day 2 closes with a team rollout plan.
Schedule
Times are indicative. Layers 0–6 form one graph — each module builds on the previous. Day 2 ends around 17:00 at a standard group pace.
Day 1 · Fundamentals, account, communication, surfaces
9:00–17:00
-
Layer 0 · Fundamentals
What an LLM is and how an answer is produced. Model limits: hallucinations, knowledge cutoff, context window. When the model helps vs. fabricates — the programme root without which model testing makes no sense.
-
Module 1 · Account and privacy
Accounts and licensing: Enterprise (Team) vs individual (Pro/Max). Privacy settings. Training use, data retention, zero data retention (ZDR). Team admin and roles for organisations.
-
Module 2 · Talking to the model (part 1)
Basics of communicating with Claude models. Model choice and differences: Opus, Sonnet, Haiku — testing on the same tasks. Prompt structure: context, task, format, examples.
-
Module 2 · Talking to the model (part 2)
Iteration and improving answers. Custom instructions and styles (user preferences). Effort control — when to raise it, when a faster model is enough.
Day 2 · Surfaces, integrations, data, responsibility
9:00–17:00
-
Module 3 · Claude surfaces
Chat, Cowork, Design, Code — split and differences. **Decision map: which surface for which task** (key slide for designers and team leads). Claude in Slack (Claude Tag) for team workflows.
-
Module 4 · Durable context and extensions
Projects. App connections (Connectors, MCP): Gmail, Drive, Calendar. Skills. Memory — separate from Projects. Artifacts. Plugins and marketplace.
-
Module 5 · Working with data
Uploading and analysing documents. Creating files: docx, xlsx, pptx, pdf. Web search and deep research — when to enable, when not to.
-
Module 6 · Responsible use
Protecting sensitive client data. Verifying outputs and citing sources. Usage limits and costs. Team rollout plan: what to deploy in 30 days vs. later. Q&A.
Tailor your enquiry (Claude + MCP)
Helps us prepare the workshop for your licence, team profile and modules to deepen.
Licence context
Participant profile (multi-select)
Your choices are attached when you send the enquiry.
Send this configuration with your message — we will respond with next steps.
Module graph (Layer 0 → 6)
The programme follows a dependency graph:
- Layer 0 — LLM fundamentals (root)
- Module 1 — account, licensing, privacy
- Module 2 — communication and model testing
- Module 3 — Claude surfaces + decision map
- Module 4 — Projects, MCP, Skills, Memory, Artifacts, Plugins
- Module 5 — documents, files, research
- Module 6 — responsible use, costs, sensitive data
Without Layer 0, participants do not know what they compare in Opus vs Sonnet vs Haiku tests. Without Module 1, company data may land in the wrong settings. Module 3 gives one map for the whole team — instead of “everyone stays in chat”.
Areas
- AI & transformation