Service 04

AI Training

On-site workshops, 2 days, tailored to your challenges.

Modalities
  • On-site, 2 days (your premises or partner venue)
  • €2,800 excl. tax + travel and accommodation depending on country
  • Group of 4 to 12 people
  • Learning materials provided (slides, exercises, references)
5 training themes

Introduction to agentic systems

What an AI agent is, how it perceives its environment and acts. Difference between a bare LLM, a tooled agent, and a multi-agent system. Concrete cases: email triage agent, document search agent.

  • Resource → What is an agent
  • Resource → LLM vs agent
  • Resource → Anatomy of an agent

Operational prompt engineering

Few-shot, chain-of-thought, structured outputs with Zod or Pydantic. Guardrails and output validation: how to detect and correct hallucinations before they reach production.

  • Resource → Prompt engineering basics
  • Resource → Few-shot learning
  • Resource → Guardrails and validation

Agent graph architecture

LangGraph: nodes, edges, shared state, cycles and correction loops. When to use a graph vs a linear pipeline. Design patterns: supervisor, subgraph, agentic map-reduce.

  • Resource → LangGraph introduction
  • Resource → Graph design patterns
  • Resource → When to use LangGraph

AI integration into existing systems

Adapters, interface contracts, swapping external tools without breaking the rest. ERP/CRM connections, credential management, auditing the existing system before automating.

  • Resource → AI adapter pattern
  • Resource → ERP/CRM connection
  • Resource → Agent credential management

LLM cost monitoring and optimisation

Langfuse: traces, dashboards, alerts on nodes that cost the most or fail the most. A/B testing prompts, routing Haiku/Sonnet by task complexity, cost reduction without quality loss.

  • Resource → Langfuse getting started
  • Resource → A/B test prompts
  • Resource → Haiku/Sonnet routing

Organise a training ?

Describe your team and objectives. We propose a tailored programme.

Describe my project →