< Back to AI Lab Hub

Follow-up session I

Thursday, March 12 from 12:30-2:00pm via Zoom

Overview

This session covered three areas: a deeper dive into AI Assistants, an introduction to AI agents, and questions and predictions for the future.

Quick Links
🌟
Key points
💻 Session slides
📺 Session recording
💭
Workout
📚 Further resources
🧪 Your AI Lab Project brief

Key points


  • This session covered three areas: a deeper dive into AI Assistants, an introduction to AI agents, and questions and predictions for the future.

  • The number one tip for AI Assistants: spoon feed them. Make your instructions almost painfully detailed and specific.

  • Think about guardrails — what do you NOT want it to do? What's out of scope? Add that to your instructions.

  • Narrow scope = better. Be clear about what details your assistant needs before it can get going, and state what reasonable assumptions you want it to make.

  • You can use shortcuts to make one AI Assistant multipurpose

    • Example from my Email Assistant instructions:

      ##SHORTCUTS

      Anytime you see the following two letter shortcuts precede a message from me, you are to ignore the above instructions and use these new instructions instead.

      “SL”: Write a short, Slack-style message with Fiona’s tone, informal, 25-125 words, no paragraph breaks, lots of in-line emojis (4+).

  • AI Assistants may be imperfect, but they build the agentic mindset we all need.

  • AI agents are different from assistants — they do work for you proactively, not reactively. They can take actions autonomously, coordinate tools, and work solo or together like a team.

  • Key agentic concepts: multi-agent systems (groups of specialised agents collaborating), human in the loop (keeping a human involved to check the work), and vibe coding (using AI to code by describing what you want in plain language).

  • Every agent flow has the same raw ingredients: a task, a desired answer, a model (the AI brain), and tools it can access.

  • There's a lot of hype around AI agents right now. They're truly like babies at this stage — but they will improve fast and will revolutionise how we work.

  • Big open questions remain: how much control do we give agents, what access and rights will they have, and how do we verify they're getting it right?

Session slides


Workout

AKA your post-session actions to take

From here on out, the AI Lab Project is your focus.

Find the full project brief here – deadline is March 20th.

Further resources

📌 Your AI Lab Project Brief in full

AI Assistant Instructions Appendix

COIF Prompt Writer GPT
This free, open source AI assistant will write fabulous COIF prompts for you with minimal instructions

Markdown Guide
To help you build your AI assistants in Markdown for optimum results