Prompt Engineering

Summary

This second session, led by Mollie, focused on prompting, artifacts, and memory. Mollie showed how to write effective prompts by giving clear context, a specific task, and constraints, using weak and strong examples to make the point. She explained that artifacts are content created outside the chat, such as documents, tables, and spreadsheets, which can be downloaded or shared with colleagues, and demonstrated how to manage them from the left-hand sidebar.

The session also covered memory, which stores recurring context about you and your preferences, along with searching past chats and importing memory from other providers. The closing point: the best way to improve at prompting is to keep working with Claude directly.

Key takeaways

• A good prompt has three parts: clear context, a specific task, and constraints.
• Comparing a weak prompt with a strong one shows why specificity matters.
• Artifacts are content created outside the chat, including documents, tables, and spreadsheets.
• Manage and revisit artifacts from the left-hand sidebar.
• Memory stores recurring context; you can search past chats and import memory from other tools.
• You learn prompting fastest by practising with Claude itself.

Chapters

1. 00:00Large language models, in context
2. 03:04Prompting techniques
3. 12:56Artifacts: the concept
4. 18:02Managing and sharing artifacts
5. 22:45Context and memory management
6. 33:48The memory feature
7. 38:04Company-level personalisation
8. 41:39Prompting best practices

Resources

• Recording, slides, transcript & chat 8 files on Zoom→