How to capture without an app

The fastest way to capture a plan is to never open a blank form. Composed gives iPhone three capture modes — voice, screenshot, or one typed line — and Apple’s Speech framework plus the parse-event service fill in the date, time, place, and AI prep checklist so you don’t tap through six fields. The whole point is to remove the moment where you stare at an empty event screen and decide it’s not worth it.

Most planning systems lose the game in the first ten seconds, before any feature even gets a chance to help. You remember the orthodontist wants a follow-up in six weeks. The window to act is the length of a hallway. If capturing it means opening an app, finding the plus button, typing a title, scrolling a date wheel, and picking a time, you will not do it — you’ll tell yourself you’ll remember, and the thought will dissolve before you reach the parking lot.

The 10-second rule

The 10-second rule is simple: whatever you just remembered has to be inside a system you trust before the next interruption arrives. Ten seconds is roughly how long a fresh intention survives in working memory while you’re mid-stride, mid-conversation, or mid-traffic-light. After that, the next thing — a text, a kid’s question, the light turning green — overwrites it.

This is not a discipline problem. It’s a design constraint. The brain didn’t evolve a notification queue. The Tuesday you stood at the pediatrician’s checkout desk, the receptionist said “we’ll need to see her again before school starts,” and you nodded — you genuinely meant to add it. By the time you’d buckled the car seat, it was competing with lunch, the grocery list, and a work Slack, and it lost.

So the design question is not “how do I remember better.” It’s “how do I make capture faster than forgetting.” Ten seconds is the budget. Every capture mode in Composed is built to come in under it.

Capture is the only step in planning that’s truly time-sensitive. Everything else — the prep, the reminders, the leave-by math — can happen later. But if the thing never gets in, none of the rest of the system ever sees it.

Voice: the fastest mode

Voice is the fastest mode because speaking a full sentence takes about two seconds and your hands stay free. You hold the mic, say “orthodontist follow-up for Maya, six weeks from Thursday, that office on Cedar,” and Composed transcribes it with Apple’s on-device Speech framework, then extracts the title, the date, the time, and the place into a real event. No typing, no date wheel, no app-switching.

The reason voice wins for capture specifically is that it matches the moment capture happens in. You remember things while walking to the car, leaving a meeting, or standing in the school pickup line — exactly the moments when typing is awkward and looking down at a screen is rude or unsafe. Voice closes the gap without asking you to stop what you’re doing. The full method for phrasing a voice capture so it lands cleanly the first time is the next chapter, and the feature itself lives at voice input.

Screenshot: the no-typing mode

Screenshot is the no-typing mode for plans that already exist as an image. A flight confirmation, a hotel booking, a birthday-party invite that came as a flyer, a school calendar someone texted you — you don’t retype any of it. You screenshot it, hand the image to Composed, and Claude reads the details and builds the event.

This is the mode for the plans that arrive already-formatted but in the wrong place. The airline emailed your itinerary; it’s sitting in your inbox doing nothing useful for your Thursday. The dance studio posted the recital date as a graphic on Instagram. Re-typing that into a calendar is exactly the friction that makes people skip it. A screenshot is one button. The full set of inputs that extract well — and the edge cases that don’t — is covered two chapters on in screenshot anything with a date.

Type: the quietest mode

Typing is the quietest mode, and it earns its place precisely when voice and screenshot don’t fit. You’re in a silent waiting room and can’t speak. There’s no image to screenshot — the plan exists only in your head. You’re somewhere a voice capture would feel performative. So you type one line: “lunch with Dana next Friday at noon, the place on Main.” One line, one field, not six.

The distinction that matters is that Composed’s typed capture is still one sentence into one box, not a form. You’re not picking a date from a wheel and a time from another wheel and a location from a search field. You write the way you’d say it, and the same parse-event extraction that handles voice handles the typed line. The mode changed; the friction didn’t come back. Typing is the fallback that keeps the 10-second rule intact when the room is quiet and there’s nothing to screenshot.

When to use which

The decision is faster than reading it: speak when your hands are busy, screenshot when the plan is already an image, type when you can’t make a sound. Here’s the matrix, in the order you’ll actually hit them.

  • Walking, driving, holding a kid, mid-transition — voice. Two seconds, hands free, eyes up.
  • It arrived as an email, a flyer, an itinerary, an invite graphic — screenshot. Don’t retype what a camera can read.
  • Silent room, nothing to screenshot, plan is only in your head — type one line.
  • You have time and want to add prep or detail by hand — type, then open the event and edit. The capture and the refinement are two separate moments.

None of these modes count against anything for plans you pull in from your calendar, by the way: Apple Calendar and Google Calendar events imported into Composed are excluded from the free-tier event cap, so syncing your existing schedule never costs you a slot.

The habit to build this week is reflexive mode-selection. Next time you remember something, don’t decide whether to capture — that decision is what eats the ten seconds. Decide only which mode the moment allows, and capture in that one. Voice if you can talk, screenshot if it’s on screen, type if it’s neither. Pick the mode the situation hands you, and let the AI do the structuring.

Next: Voice to calendar — the method — how to phrase a voice capture so it resolves correctly the first time, and what to do when it doesn’t.