Tag: Engineering & Tools

All the articles with the tag "Engineering & Tools".

• Checking After Hitting Send

  9 Oct, 2026

  When an AI agent rethinks its response after hitting send, it is behaving more like a human would. Why this also points to a co-dependency between harness and model.

• Why Trust & Safety is Becoming an Agent Platform’s Top-Line Problem

  2 Sep, 2026

  Automated moderation models that over-flag developers testing AI agents aren't just an annoyance — they directly risk developer retention and top-line platform revenue.

• When Ground Truth Gets Dirty

  23 Jul, 2026

  AI is a powerful exploration tool — but not ground truth. As trusted sources start generating docs with AI, errors compound through summaries, memory layers, and agents. HITL isn't optional.

• The Assistant Was a Phase

  25 Jun, 2026

  A deadpan walk through vibe coding, context engineering, harness engineering, and loop engineering — ending at the obvious conclusion that nobody wanted to say out loud: it's not an assistant anymore.