Weekly Roundup #28 — Claude Models, TLS, and What Worked

This week was all about Claude models coming back and TLS finally getting its due on the blog. I published 6 posts between Monday and Friday, covering everything from Anthropic’s redeployment of Fable 5 to a deep dive on why SSL is dead and TLS runs the web. It was a dense week, and the numbers reflect that.

By the numbers: 6 posts published across 3 categories. The Claude Sonnet 5 piece led in page views with 9, followed closely by the SSL/TLS explainer at 8. But the real standout was Claude Fable 5 Is Back — only 6 views, but 109 seconds average time on page and a 50% bounce rate. That tells me people who clicked actually read it. The TLS 1.3 features post and the 60% API caching story rounded out the week with 3 views each.

Best-performing post: Claude Fable 5 Is Back. The engagement numbers are what caught my attention — 109 seconds average duration is the highest I have seen in weeks. The takeaway: timely news analysis with clear context (what happened, the catch, new safeguards, why it matters) keeps people reading. I structured it as a timeline from June 9 to July 1, and that narrative flow worked. The lesson for me is that breaking AI news with a structured breakdown — not just reporting the fact but explaining the implications — drives real engagement even with modest traffic.

What I learned: The TLS/SSL posts (SSL Is Dead and TLS 1.3 Features) both landed well in search impressions. The GSC data shows these are starting to get traction, which makes sense — it is evergreen infrastructure content that people search for year-round. I am going to double down on this format: foundational tech explained clearly, with a practical angle.

What is next: I am working on a piece about running local LLMs on consumer hardware — the Lenovo AI Node experiment I have been tinkering with. Also planning a follow-up on the Claude ecosystem now that Fable 5 is fully back, specifically comparing its agentic capabilities against Sonnet 5 in real coding workflows. And I want to revisit the API caching story with actual benchmarks from the past week.

That is week 28. See you next Monday.