Workshops: Prompt Engineering

Advanced Prompt Engineering Workshop: Beyond the Basics

This workshop is for people who know the basics and now want to solve demanding tasks reliably with AI. You will learn few-shot prompting, role and system instructions, structured output and how to work with reasoning models, plus when each technique actually makes a difference.

Who is this workshop for?

  • Power users who already work with ChatGPT, Claude or Copilot daily
  • AI champions and multipliers in business units
  • Analysts, consultants and knowledge workers with complex tasks
  • Teams that feed AI output into spreadsheets, tools or workflows
  • Graduates of our basics training

What you will take away

  • You use few-shot examples so that style and format are adopted consistently.
  • You produce structured output such as tables, JSON or fixed templates that can be processed directly.
  • You know when reasoning models are worth it and how prompting them differs from classic chat models.
  • You break complex tasks into prompt chains with clear intermediate results.
  • You use custom instructions, Custom GPTs, Claude Projects or Gems for recurring work.
  • You detect and reduce hallucinations through source grounding and self-checks.

Agenda

  1. Module 1

    Quick check: where are we?

    • Review of prompts brought by participants
    • Why good prompts still fail: ambiguity, missing context, conflicting goals
  2. Module 2

    Examples, roles and system instructions

    • Zero-, one- and few-shot compared side by side
    • Using roles purposefully instead of role folklore
    • Persistent instructions in Custom GPTs, Claude Projects and Gemini Gems
  3. Module 3

    Structured output

    • Tables, Markdown templates, JSON and XML tags
    • Enforcing and validating output formats
    • Exercise: turn unstructured documents into a data table
  4. Module 4

    Getting the most from reasoning models

    • What reasoning models do differently, and what they cost
    • Goals and criteria instead of step-by-step instructions
    • Exercise: the same analysis with a chat and a reasoning model
  5. Module 5

    Prompt chains and quality assurance

    • Splitting tasks: research, draft, critique, revision
    • Self-checks, source quotes and cross-checking with a second model
    • Building your own multi-step workflows

From good results to reliable results

Writing one good prompt is easy. It gets hard when the result must have the same structure, the right tone and solid substance every single time, for reports, analyses or content that flows into other systems. That is where advanced techniques matter.

We show not only how the techniques work but also when they are unnecessary. Modern models need fewer tricks than earlier ones, but they need precise context and clear success criteria more than ever.

Models and tools compared

Different models respond differently to the same instruction. In the workshop we compare ChatGPT, Claude, Google Gemini, Microsoft 365 Copilot and Mistral Le Chat, including their reasoning variants, on identical tasks. You develop a feel for which model suits which job.

Data protection in advanced use

Uploading documents, creating projects with knowledge bases or piping output into other systems means more responsibility. We discuss which content belongs only in business plans, how to handle personal data under the GDPR and which internal approvals make sense.

Frequently asked questions

How does this differ from the basics training?

The basics training covers prompt structure and safe use. The advanced workshop assumes that knowledge and focuses on techniques for consistent, structured and complex results.

Do participants need to code?

No. We explain structured formats like JSON without programming. Anyone who wants to integrate AI into software via API is better served by our workshop for developers.

Can we bring our own tasks?

Yes, that is strongly encouraged. We ask for anonymised examples in advance so we can tailor the exercises to your subject matter.

Workshop

Get in touch

Ready to take your power users to the next level?

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?

We only use your details to handle this request. See our privacy policy.