All Articles

The SpaceX acquisition of Anysphere became effective on August 14. Origin's early beta followed on August 17. Reading the gap from the filing itself.

Git history is the cheap part of a code-host move. The bill sits in runners, packages, deploy trust, branch rules and webhooks. An inventory-first framework.

An agent that always replies instantly pushes first-response time near zero and stops it separating good service from bad. A replacement metric set instead.

GitHub's August 12 changelog puts Agent Plugins 1.0 support at general availability. What ported across harnesses, and what a plugin still cannot carry.

Claude Code shipped four tagged releases in roughly 50 hours, one switching a tool family off by default, reversibly. Running a team on a shifting harness.

Anthropic's engineering post publishes auto-mode classifier results across three separate datasets. Why an FPR and an FNR from different sets cannot be paired.

Credentio validates C2PA Content Credentials on-device, and Google says it runs in nearly 40 of its own products. What a credential proves, and what it cannot.

Google added Gemini 3.7 Flash as an AI Mode option for Pro and Ultra subscribers, in English, with no ranking change. What SEO teams should measure instead.

From August 24 Google moves commission-eligible YouTube affiliate clicks out of Merchant Center organic traffic, and restates history. Not lost demand.

Press reported that investors are targeting a $2 trillion Anthropic IPO, a figure the company has not fixed. A reading protocol for numbers that arrive by leak.

A census of 30 models publicly described as open in 2026, with the licence read directly wherever the tag was non-standard. Full table, method and caveats.

We collected 42 vendor-published benchmark rows from 9 vendors and checked what each page discloses: version, subset, harness, effort. Full table in the post.

Field, note, task or activity log: an agent has to choose a write target. A framework that keeps reporting fields clean and context where humans read it.

Price-change notice is a contract term, not a courtesy, and the periods range from a fortnight to none at all. What to check before committing model spend.

Most AI caption failures are voice failures, not grammar failures. A guardrail system covering brand voice, platform mechanics, and feed-level repetition.

Three Chinese labs shipped major releases and their licence stories diverge. A buyer's scoreboard on weights, pricing schedules, and benchmark claims.

CRM timeline and field-history APIs already record who changed what, when, and from what. How to turn that trail into audits agents run and humans approve.

Introductory Gemini Flash pricing is published through 31 December, then steps up. What paid-media teams should rewire and move to batch before it changes.

A vendor-published disclosure ledger reports thousands of findings across hundreds of open-source projects. What maintainers and security teams should check.

Every vendor names its reasoning-effort settings differently and the rungs do not line up. What the ladders look like, and where thinking cannot be disabled.

Crossing a context threshold can reprice an entire request, not just the tokens above the line. Where the cliffs sit and how to keep prompts underneath.

Time-of-day LLM pricing is arriving, and a cheaper window is not the same as a cheaper bill. How the announced schedules and peak hours actually work.

DeepSeek open-sourced its dsh agent harness under MIT: a developer preview at 0.1.0-rc.5, four runtime modes, and every part of the product a plugin.

DeepSeek's changelog now carries a GA entry for V4-Pro across app, web, and API, MIT-licensed 0813 weights on Hugging Face, and pricing changing August 16.