Typing works fine when you're already sitting at a keyboard, already thinking in text, already ready to type a full sentence. A lot of the moments where you'd actually want help aren't like that, you're mid-task, hands on something else, or the request is short enough that opening a window and typing it out feels like more overhead than the task itself. That gap is why Alden treats voice as the default interface, not an add-on.
Alden listens for a wake word rather than staying in an always-transcribing state. Say it, and Alden is listening; the rest of the time, it isn't actively parsing your speech into a request. That distinction matters both for how the interaction feels (you decide when it listens, and nothing runs in the background) and for keeping the experience predictable: you know when Alden is engaged and when it isn't.
A lot of voice assistants still work in a record-transcribe-respond loop: you speak, it processes, then it replies, with a noticeable gap in between. Alden runs as real-time speech-to-speech instead, so the back-and-forth feels closer to an actual conversation than a queue of commands. That immediacy is what makes voice a genuinely usable primary interface rather than a novelty layered on top of a text-first product.
Because voice is the primary way Alden shows up, how it sounds matters as much as what it says. Alden's voice is calm, precise, and concise, with a quietly confident tone and a restrained dry humor: its own presence, deliberately not modeled on any fictional AI assistant. That was a conscious choice, not an accident. A voice-first assistant that operates your computer on your behalf should feel trustworthy and specific, not like a reference to something else.
Voice is the fast path in, but it isn't the only surface. Alden's Workspace gives you a text-based view of the same conversation and memory for anything that benefits from being written down, reviewed, or revisited, we cover that split in more detail in the Orb and the Workspace. And because Alden can act on your behalf, filling forms, managing files, running workflows, voice being fast doesn't mean it's unchecked: every action still routes through the same Governance Engine described in what is governance-first AI, whether you asked for it out loud or typed it.