Sure, AI is pretty cool, but have you heard of something called Retrieval-Augmented-Generation (RAG)... We don't often spotlight specific tech on the Show, but RAG?
I firmly believe that RAG has the potential to shake up service design in a big way.
Imagine having a super-powered teammate on every project. This teammate has the ability to recall every meeting, every workshop, and every sticky note, not just yours but your entire team's, even from years ago. Not just yours but your whole team's.
Ask them a question, and a few seconds later, they've got the answer. It's like being able to have a conversation with your entire project history. Just think about the impact of this for a moment.
Now, we all know about those fancy Large Language Models (LLMs) like ChatGPT. Amazing, right? But they're not trained on your data. Ask them about your project, and you'll get... well, something made up. But what if you could combine the conversational magic of LLMs with the deep knowledge of your own data?
In a nutshell, this is RAG's promise. It lets those powerful LLMs tap into your world, giving you answers that are not only smart, but relevant.
I've been tinkering with RAG to unlock the wisdom hidden in our Circle community discussions. But I'm far from an expert, so I brought in someone who is: Kirk Marple, founder of GraphLit, a startup using RAG to make your knowledge AI-friendly.
In our conversation we dove deep. How do we even start with RAG? Do you need to be a coder? How do we make sure the answers you get are any good? What about privacy when AI sees your data? And that's just the start to be honest.
What struck me was Kirk's idea that using AI is more art than science. It's about 'prompt sculpting', not (just) engineering. There's a lot of gray area, and that's where we as a design community shine.
We should be all over this AI thing... What do you think?
--- [ 1. GUIDE ] ---
00:00 Welcome to Episode 209
05:00 What Kirk does in life
10:00 AI for content discovery
14:00 AI and service design
16:00 Data retrieval with AI
19:00 Tracking unstructured data
22:00 Podcast metadata example
24:30 Vector search explained
30:00 AI vs human experience
35:00 Privacy concerns with AI
37:30 Large language models and understanding
41:00 Importance of graphs in AI
44:30 AI: art or science?
48:00 AI's growth and data processing
51:30 AI agents
56:00 Kirk's AI roadmap
57:30 Tips for AI beginners
59:00 Common AI terms
1:01:00 AI resources
--- [ 2. LINKS ] ---
--- [ 3. CIRCLE ] ---
Join our private community for in-house service design professionals.
https://servicedesignshow.com/circle
