Intelligence isn't just about the model When we talk about AI products, we usually compare models. Qwen vs Gemini. Claude vs GPT-6. Reasoning vs Non-Reasoning. Latency vs Context window vs Benchma
blog.mdazlaanzubair.com3 min read
Auto-focus fits the explicit Ask Gemini action, but it should come with a predictable focus contract: announce the panel to assistive technology, place focus in the prompt, and let Escape return it to the previously selected page element. A keyboard shortcut that opens and focuses the panel would make the intent even clearer than a context-menu path. The useful metric is not clicks alone but time-to-first-input, focus errors, and how often users type into the page by mistake.
The mismatch between "visual readiness" and "interaction readiness" is a sharp way to name this, and it generalizes past Gemini specifically, a lot of "this feels sluggish" complaints trace back to exactly this gap, where the UI signals done loading while some other layer (focus, hydration, a pending API call) hasn't actually caught up yet. Users don't distinguish between "the model is thinking" and "the interface isn't ready," they just experience both as the same kind of friction.
The focus-contract discussion in the comments is the more rigorous version of this observation, and worth sitting with: auto-focus without an escape path just relocates the friction instead of removing it, turning a one-click annoyance into a keyboard trap for anyone navigating without a mouse. That's the part easy to miss if you're only testing the happy path yourself, the fix that feels obviously right for one interaction mode (click, type immediately) can be actively worse for another (keyboard-only, screen reader) unless the whole cycle, open, focus, escape, restore, is treated as one contract instead of a single click removed.