Workshops: AI Automation & Agents
Workshop: AI Workflows with Claude, OpenAI & APIs
In this two-day workshop engineering teams learn to integrate AI models into their own applications and workflows via APIs – with function calling, structured outputs and the Model Context Protocol. The focus is on robust, secure and cost-aware implementations.
Who is this workshop for?
- Software and backend teams integrating AI features into products
- Automation specialists hitting the limits of no-code tools
- Technical architects planning LLM integrations
- Data engineers with Python or TypeScript skills
- IT teams opening internal systems to AI agents
What you will take away
- You call Claude and OpenAI models via API and control parameters, context and streaming.
- You define tools with clean schemas and let models call your own functions.
- You enforce structured outputs and validate them before further processing.
- You build an MCP server that exposes internal data to Claude and other clients.
- You implement error handling, retries, caching and cost control.
- You secure tool calls against prompt injection and unintended actions.
Agenda
Module 1
Day 1: API fundamentals
- Anthropic Messages API and OpenAI APIs compared
- System prompts, context windows, tokens and cost
- Streaming and prompt caching
Module 2
Day 1: Function calling and tool use
- Defining tools: names, descriptions, JSON schemas
- The tool loop: call, execute, return
- Exercise: an assistant that queries and updates CRM data
Module 3
Day 1: Structured outputs
- JSON schema constraints and validation
- Extraction and classification pipelines
- Evaluation with test datasets
Module 4
Day 2: Model Context Protocol
- MCP architecture: servers, clients, tools, resources
- Building your own MCP server
- Using it in Claude, Claude Code and other clients
Module 5
Day 2: Production readiness
- Error handling, rate limits, fallback models
- Logging, monitoring and cost budgets
- Security: permissions, confirmations, input validation
- Developing faster with Claude Code or Codex
Why the API rather than chat interfaces?
Chat interfaces are ideal for individuals. As soon as AI needs to be embedded in processes, products or large data volumes, the API is the way to go. You decide on model choice, data flow, logging and cost.
Function calling turns a language model into an orchestrator: it decides which of your functions to call with which parameters and continues working with the results.
Across models and vendors
We work primarily with Claude and OpenAI models but show patterns that also work with Gemini or Mistral, so you avoid tight vendor lock-in.
- Abstraction layer for multiple providers
- Model choice per task: quality, latency, cost
- EU hosting options via cloud platforms
- Local models for highly sensitive data
Security for tool calls
A model that may call functions can also be tricked into calling the wrong ones. We cover prompt injection via tool results, per-tool permission concepts, confirmation steps for write actions and secure API key management. Our AI Code Engineers design the exercises from project experience.
Frequently asked questions
Which programming language do we use?
Python or TypeScript by default. We adapt the exercises to your team's stack.
Do we need our own API access?
Yes, the exercises require API keys from Anthropic and OpenAI or via a cloud platform. We help with setup beforehand.
How is this different from the n8n workshop?
The n8n workshop works visually with ready-made nodes. This workshop is for developers who integrate AI directly in code and need full control.
Workshop
Get in touch
Should your engineering team integrate AI features securely and production-ready?
Tell us a little about your team – we will get back to you within 24 hours with a proposal for content, format and dates.
Prefer to talk?
More workshops
- Workshops: AI Automation & AgentsWorkshop: Automate Social Media & Content with AI
- Workshops: AI Automation & AgentsAI in Project Management: Jira, Asana, Notion & Copilot
- Workshops: AI Automation & AgentsIdeation Workshop: Find & Prioritise AI Use Cases
- Workshops: Prompt EngineeringAdvanced Prompt Engineering Workshop: Beyond the Basics
