Integration of AI into the software development process has been widely embraced, but the proper workflow has been under debate. Now a consensus is emerging that there should be an executable spec, separate from a prompt, markdown doc, or tests. Perhaps even the AI should not have access to the spec.

The team behind Nexter has created 1,050 ready-made AI prompts (recipes) for their 105 blocks. To quote, “Copy a recipe, paste it into your AI assistant, and it builds a real Nexter block in the editor. The Recipe Library has ten prompts for each of 105 blocks and works with any assistant that supports MCP, like Claude or Cursor.”

It looks like the AI MCP is going to be integrated into core. There is some discussion about this.

To quote, “We’ll use it to find and fix vulnerabilities across all our products, from the free version to SEOPress PRO, and to prevent incidents.”

An interesting change in perspective – Kevin Powell suggests that companies were very happy to embrace AI when it meant firing their employees, but what happens when people replace the companies software with AI created clones?

Something we’ve all seen is the increase of a platform providing more or different features on the one hand and for customer lock in on the other. This article breaks it down.

Kevin Powell shares what people are saying in conference hallway chats and at companies about the sift to 100% AI. He echoes what Tobi Salami said about creating tutorials, is there still the need?

What Jordy Meow has been doing with AI is way cool. His AI Engine plugin is very advanced. I’ve been trying the AI chat plugins that developers put on their sites for support. His, Nyao, is very thorough and gave me information way beyond what’s in the docs. Now he has added free Android and iOS apps to use the AI features from your phone.

I can’t help but think this article is aspirational and not aligned with reality, but it suggests the direction in the age of AI.