How This Works

Most blogs are a list of posts in reverse chronological order. Write something, publish it, move on. This one works differently.

Two types of content

Signals are living documents. Each one tracks a topic that doesn’t have a resolution date – the state of AI, the engineering jobs picture, the agentic infrastructure story. They get updated regularly as new evidence comes in. The date you see on the index is when it was last updated, not when it was first published. Every update creates a dated snapshot so the history is preserved and linkable.

Articles are one-off pieces. A specific incident, a specific decision, a specific piece of research worth examining properly. Published once, fixed in time.

AI-assisted production

Every post here is researched and written with AI. That’s not a disclaimer – it’s part of what makes this format possible.

The production process: I set the brief, the angle, and the editorial judgment. AI does the research, synthesis, and first draft. I review, correct, and approve what goes out. The curation is human. The labour is shared.

The alternative – writing everything from scratch – would produce one post a month, not the coverage this format requires. The Signals model only works if updates are genuinely frequent. AI makes that viable.

The disclaimer on every post – Commissioned, Curated and Published by Russ. Researched and written with AI – is the honest summary of how this works.

Why Signals

Engineering and AI are not static topics. A post about the state of AI written in February is outdated by March. The standard blog response is to write a new post. The problem: the new post exists in isolation, the old one is stale, and readers have no single place to track a topic over time.

Signals solve this. One canonical document per topic, kept current, with a full revision history at dated snapshot URLs. If you bookmarked the State of AI thread in February, the same URL has March’s update. The history is still there at /ai-state/20260302/.

Sources and standards

Sources are cited inline. Claims without sources are opinions and are framed as such. When the research comes back ambiguous, that’s noted rather than papered over.

The one rule: nothing gets published without a verifiable source. The AI research step is specifically instructed to discard any claim it can’t back with a real URL.

Who this is for

Senior engineers and engineering leadership who want to stay current without spending two hours a day reading. The audience knows what an API is. The posts don’t explain basics, don’t sell optimism, and don’t pretend uncertainty doesn’t exist.