Who is it for?

Professionals who already use AI assistants and want to move from casual use to method: analysts, consultants, managers, researchers, writers.

Objectives

By the end of the training, participants can:

Apply advanced prompting patterns (roles, few-shot, chain-of-thought, output formats)
Decompose complex business tasks into reliable AI workflows
Use research and document tools (deep research, document Q&A, corpus analysis)
Detect, control, and mitigate hallucinations in their daily work

Program

Day 1 — Foundations and prompting method

The mental model that makes everything click. Tokens, context window, training vs inference, where hallucinations structurally come from — just enough theory to predict the model’s behavior instead of being surprised by it.

Choosing your tools. The model landscape: Claude, GPT, Gemini, Mistral and when each fits. Chat interfaces vs APIs vs vertical tools. Reading a pricing page without being fooled.

Prompt engineering, from basics to solid. Roles (system, user, assistant) and the anatomy of an effective prompt. Few-shot examples. Chain-of-thought. Controlled output formats (JSON, markdown, tables). When extended thinking is worth the cost.

Hands-on — your prompts, audited. Each participant brings a real task; we rewrite the prompts together and measure the difference.

Day 2 (optional) — Workflows, research tools, reliability

Advanced patterns. Decomposing complex tasks into steps. Self-critique and refinement loops. Common anti-patterns that silently degrade output quality.

Research and document workflows. Deep research tools (Perplexity and equivalents), document Q&A, multi-document corpus analysis with NotebookLM, claim-checking with consensus tools. Building a literature review or competitive analysis with sources you can actually verify.

Reliability in daily work. Hallucination mitigation strategies: grounding in documents, asking for sources, cross-model checks. What to never paste into a chatbot — confidentiality rules that hold in practice.

Hands-on — build your workflow. Participants assemble a complete AI workflow for one of their recurring tasks and document it for their team.

Practical details

One or two days depending on depth, on site or remote
Participants work on their own real use cases
Includes post-training follow-up calls to review your AI projects