Editorial queue, REST versus RPC, and the agentic metaphor

The day was about turning recent research into publishable work. The focus stayed on the blog as an editorial pipeline: choosing themes, writing bilingual pairs, adjusting dates, and giving each post enough image and context to travel well.

REST, RPC, and internal systems

I published a post comparing REST and RPC, with a focus on when to use gRPC or tRPC. The point was not to create an abstract style fight. It was to explain the kind of decision that shows up in real systems: public boundary, internal contract, latency, typing, compatibility, and maintenance cost.

I also added a cover image and a simple client-server flow diagram. It became a more architectural post, useful for organizing vocabulary and showing that an API choice should follow the shape of the problem.

Agentic engineering

Another step was the post that uses an autonomous car as a metaphor for the shift from AI-assisted engineering to agentic engineering. I worked through the idea that the point is not only having a smarter tool. The engineer's role changes: from doing every step to designing limits, checkpoints, review routes, and intervention criteria.

Then I adjusted the publication date to keep the queue coherent. That small care matters when several agents are working in parallel.

References for agents

There was also research into skills, frontend quality, and ways to guide agents with clearer instructions. The pattern showed up again: better operating context reduces the chance that an agent improvises where it should follow a contract.

It was a day of aligning text, system, and process. The blog gets better when each publication also improves how the next ones will be made.