Every item below was published by Anthropic on anthropic.com, X or LinkedIn in January 2026, listed newest first with its date and a link to the original post. The launches that change something for a European business running Claude also carry a short note from us.
Anthropic redesigned its take-home test three times as Claude improved, so that interviews still say something about human ability.
Tech companies need to rethink their coding interviews, because this is no longer theory. AI handles take-homes, so the real question is what we are testing now.
Evaluating agents requires a combination of code graders, model graders and human review, for both results and interaction quality.
Evaluating agents is where we get stuck with many clients. This is a good base map. Save the link for the next proof of concept.
Analytics cookies show which pages are read. Marketing stays off unless you tick it under Details.