AI Agents, MCP and the problems with AI benchmarks with Matt Carey

Описание: In this episode, I spoke with Matt Carey, founding AI engineer at StackOne, founder of AI Demo Days and member of the OpenUK AI Advisory Board.

Everyone needs a friend who works in AI to help them filter the AI news and get the signals from the noise. Matt is that friend for me!

We discussed AI agents, MCP, and the challenges of AI benchmarks, which help explain the disconnect between the benchmark results and the anecdotal experiences of AI users, such as myself.

Links from the episode:

Google's whitepaper on AI agents https://archive.org/details/google-ai...
Anthropic Building Effective AI Agents https://www.anthropic.com/engineering...
Simon Willison on X https://x.com/simonw
Thorsten Ball's Joy & Curiosity newsletter https://substack.com/@thorstenball
AI Demo Days https://demodays.ai/
MCP has a prompt injection problem https://simonwillison.net/2025/Apr/9/...

✅ Recommended playlists:

Real-World Serverless podcast on YouTube
   • Is AWS Bedrock the OpenAI killer?  

My BEST videos of 2024
   • Is AWS Bedrock the OpenAI killer?  

🔔 For more free videos on YouTube, subscribe to my channel:
   / @theburningmonk  

💌 If you're hungry for more insights, best practices, and tips on building serverless applications, make sure to subscribe to my weekly newsletter and elevate your serverless game!
https://theburningmonk.com/subscribe

📚 My courses 📚
https://theburningmonk.com/courses

👨‍🏫 My hands-on serverless workshop 👨‍🏫
https://productionreadyserverless.com

🌐 Find me here
Twitter:   / theburningmonk  
LinkedIn:   / theburningmonk  
Blog: http://theburningmonk.com