Daily Reading List – June 20, 2025 (#572)
Yesterday was a US holiday. It’s weird having a Thursday off and then going right back to work on Friday. But, I think we all survived. Got some work done, did a Twitter thread on coding assistance, and got hands-on with some upcoming tech. Not too shabby.
[blog] How We Use AI At Pulley. Excellent look at what it means to build with AI. From tools to updated practices, there’s a lot to learn from here.
[article] MCP Needs a Security Reality Check. MCP didn’t start with a lot of mature security considerations, but they’re quickly coming around.
[blog] How to pass the Google Cloud Generative AI Leader Exam: A Study Guide. This is a new test, and it’s helpful to see how JK studied for it.
[blog] Single vs Multi-Agent System? Great piece by Philipp. There’s competing advice out there, and he spends time sharing the pros/cons of each approach.
[paper] Build the web for agents, not agents for the web. Interesting POV. This paper proposes a new interface for the web that’s specifically designed for agent consumption.
[blog] Design Patterns for Securing LLM Agents against Prompt Injections. Invest in this area of knowledge and keep track of the latest thinking of how to secure our AI systems.
[article] Vibe Coding in a Post-IDE World: Why Agentic AI Is the Real Disruption. Agreed. Agentic coding will have a bigger impact than the current generation of AI assistants.
[article] Navigating the Jump from Manager to Executive. It’s not an easy jump. Here’s some guidance as you move up the chain.
[blog] Designing an AI-Native Marketing Team: A Guide for Founders . With an embedded presentation. If yesterday’s playbook doesn’t apply anymore, how do you craft the right team?
[blog] Every service should have a killswitch. Every code path doesn’t require this, but there’s definitely merit in having a way to turn off a hot operation.
[blog] Developer’s guide to getting started with Gemini 2.5 Flash-Lite. This new member of the Gemini family looks solid. Here’s how to start using it.
[article] Designing Collaborative Multi-Agent Systems with the A2A Protocol. Here’s a helpful deep dive into agent collaboration, with a look at A2A and MCP.
Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below: