Claude Code in
production.
Not just in demos.
Most Claude Code tutorials stop at the chat interface. This full-day workshop starts where they end – building MCP servers that connect Claude to real systems, writing reliable agentic workflows, and deploying the whole thing to AWS Bedrock.
11:00–15:00 BST · Online · Live
40% off · use code JULY2026
6 seats available
Secure checkout via Stripe
What's included
You've used Claude Code.
But not like this.
These are the walls every developer hits when moving from experimentation to production.
Claude keeps losing context mid-task
You've hit the context window, or your CLAUDE.md isn't doing the work it should. Module 1 fixes this permanently.
MCP looks powerful but you can't figure out what to actually build with it
Most tutorials show toy examples. Modules 2 and 3 build a real MCP server connected to a real system – live, in the session.
Your agentic workflows work on simple tasks but break on anything complex
Tool chaining and failure handling are where agents fall apart. Module 4 covers both with production patterns you can take home.
You don't know how to take Claude out of your laptop and into production
AWS Bedrock in Module 5 – cost controls, latency, guardrails, and a live deployment by end of the session.
Five modules.
One production system.
Every module builds on the last. You write real code throughout – Claude Code handles the boilerplate, you handle the architecture decisions.
Most developers use Claude Code like a smarter autocomplete. This module reframes it as a stateful engineering collaborator — one that knows your codebase, follows your conventions, and produces output you can actually ship. We cover the mental models that separate developers who get 10× productivity from those who get frustrated.
What we cover
- CLAUDE.md – writing one that actually works for large codebases, not just toy projects
- Context management – how to structure long tasks so Claude doesn't lose the thread
- Test-driven prompting – writing prompts with expected outputs so you can verify results immediately
- The three most common Claude Code mistakes and how to avoid them
- Memory and session strategies – making Claude remember decisions across long builds
Who this is built for.
This is a technical workshop. You'll write real code throughout.
You're ready if you…
- ✓Write code regularly – any language, any stack
- ✓Understand what an API is and have called one before
- ✓Are comfortable in a terminal
- ✓Have Node.js installed (or can follow our setup guide)
- ✓Want to ship AI features, not just experiment
Not right if you…
- ✕Have never written code before – try our AI for Everyday Business Tasks course first
- ✕Are looking for a non-technical AI overview
- ✕Want a deep ML/model training course – this is application-layer engineering
Taught by someone who has shipped this.
Everything in this workshop comes from hands-on system building – not tutorials. InAgentic has deployed AI into real workflows: Azure-based form processing, Claude and AWS Bedrock integrations, MCP server development, and agentic workflows across internal and client-facing projects.
InAgentic is a UK-registered AI engineering company (INAGENTIC LTD, Company No. 16702679) based in Oxford, specialising in Claude Code, AWS Bedrock, and agentic AI systems.
Välj ditt workshopdatum.
Live online-workshops arrangerade i Londontid (BST).
11:00–15:00 BST · Live online-workshop · Valfri samarbetssession efteråt
12:00–16:00 BST · Live online-workshop · Bra överlappning för deltagare från UK, Europa och USA
11:00–15:00 BST · Live online · Small cohort (max 6 delegates)
Behöver du ett privat workshop för ditt ingenjörsteam?
inagentic.ai levererar också privata workshops för mjukvaruteam, ingenjörschefer och innovationsgrupper som utforskar AI-native ingenjörsarbetsflöden.
6 seats. Saturday 11 July.
Small cohort by design – every delegate gets real attention throughout the day.