Workshops: Prompt Engineering
Build and Maintain a Team Prompt Library: Workshop
In this half-day workshop your team builds a shared prompt library that people actually use. You define structure, quality criteria and maintenance processes, and turn the best prompts into reusable assistants such as Custom GPTs, Claude Projects or Copilot agents.
Who is this workshop for?
- AI champions and AI leads
- Team leads who want to scale AI use in their team
- Knowledge management and internal communications
- IT and digital transformation leads
- Teams that already have early prompt collections
What you will take away
- You define a structure for your prompt library that colleagues navigate intuitively.
- You create a standard prompt template with purpose, inputs, example and limits.
- You assess prompts against clear quality criteria and decide what goes in.
- You turn frequently used prompts into Custom GPTs, Claude Projects, Gems or Copilot agents.
- You set up roles and a lean process for maintenance, versioning and retirement.
Agenda
Module 1
Inventory
- Which prompts does the team already use?
- Collecting and prioritising frequent tasks
Module 2
Structure and template
- Categories by task, role or process
- The prompt template: purpose, variables, example, notes
- Exercise: standardise five team prompts
Module 3
From prompt to assistant
- When a prompt is enough and when a dedicated assistant is better
- Custom GPTs, Claude Projects, Gemini Gems and Copilot agents compared
Module 4
Maintenance and governance
- Owners, review cadence and feedback loops
- Where to store it: intranet, SharePoint, Notion or in-tool
- Confidentiality and approvals
Why most prompt collections die
Many teams start a list of useful prompts, and a few weeks later nobody uses it. The reasons are nearly always the same: no clear structure, no quality control, no owners and prompts that do not work without context.
A good prompt library is more like a small product than a document. It needs users, maintenance and continuous improvement, and that is what we set up in the workshop.
Platforms for shared prompts
Most business editions of the major assistants offer ways to share prompts and assistants across a team. We show the options and their limits.
- ChatGPT: Custom GPTs and shared projects
- Claude: Projects with knowledge base and instructions
- Microsoft 365 Copilot: agents and prompt gallery
- Google Gemini: Gems
- Mistral Le Chat: agents and libraries
Security in shared assistants
Shared assistants often contain internal documents as a knowledge base. We clarify who gets access to what, which content must stay out and how to limit prompt injection via uploaded files.
Frequently asked questions
Which tool is best for a prompt library?
It depends on your AI platform. A combination often works best: the library in your intranet or wiki, the most-used prompts as assistants directly in the AI tool.
How large should a prompt library be?
Start small: 15 to 30 genuinely good prompts for frequent tasks are worth more than hundreds of unchecked ones. The library then grows with actual demand.
Do we need the basics training first?
It helps but is not required. Participants should use AI assistants regularly so they can bring their own prompts.
Workshop
Get in touch
Shall we build a prompt library with your team that actually gets used?
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: Prompt EngineeringPrompting Image AI: Midjourney, ChatGPT, Firefly & Gemini
- Workshops: Prompt EngineeringContext Engineering: Prompting for AI Agents Workshop
- Workshops: Prompt EngineeringMeasuring Prompt Quality: Evaluation & Testing Workshop
- Deploy, Convert & ScaleScaling a Lovable App: From Prototype to Production-Grade
