Spotlight vs Terminal

BEAM has two input surfaces. They look similar — a place to type, an AI on the other end — but they are built for different moments, and the difference is enforced by architecture, not convention.

Terminal: the full surface

The terminal at the bottom of the screen is where BEAM's whole command set lives: every task verb, every AI mode, every setting. It's also where mutations happen. When the AI proposes an action there — creating tasks from a plan, merging docs, applying a triage pass — the terminal is where the proposal renders as a receipt and where your y executes it.

The terminal is a place you work in. Sessions have history, AI answers stream into a scrollback you can search and print, and multi-step exchanges build on each other.

Spotlight: fast in, fast out

Spotlight (Cmd+K, or ? in the terminal) is the overlay for the moment you need one thing: where a doc went, what's due Thursday, a quick answer against your workspace. Type to search and ask; prefix with / to synthesize across your workspace with sources cited.

Spotlight is read-only by construction. Its AI runs with no mutation tools attached — not instructed to decline changes, but structurally unable to make them. Ask Spotlight to do something and it will point you to the terminal. When it synthesizes, it labels what it drew from, and when your workspace doesn't contain enough to answer, it says so instead of improvising. The result is a surface you can trust reflexively: nothing you type into Spotlight can change your data.

The heuristic

Reaching for a result → Spotlight. Doing work → terminal.

If the question fits in one breath and the answer ends the interaction — Spotlight, from anywhere, without leaving what you were doing. If the answer will lead to a follow-up, a decision, or a change to your workspace, start in the terminal so the whole exchange lives in one place.

See also: Reference: Keyboard shortcuts · spotlight