How Natural Language Makes Adding Tasks 10x Faster

Creating a task should take seconds, not taps. Type what you mean in plain English and let the app figure out the rest.

The problem with traditional task creation

In most task apps, creating a fully specified task is a multi-step process. You tap "new task," type the title, tap to set a due date, scroll through a date picker, tap to set a priority, select from a dropdown, tap to assign it, pick a person, and finally save. That is six or more taps and several seconds of navigation for what is fundamentally one thought.

When something is tedious, people stop doing it. You end up with vague tasks like "groceries" with no due date, no priority, and no assignee. Which defeats the purpose of having a shared task list in the first place.

The more friction you add to task creation, the less likely either partner is to externalize their mental load into the app. And if tasks do not make it into the app, the app is not helping.

The natural language approach

Natural language task creation flips the process. Instead of filling out a form, you type (or dictate) a single sentence that describes the task the way you would say it out loud:

One line of text. One tap to confirm. The app parses the date, priority, and assignee automatically and shows you a preview with chips so you can verify everything was understood correctly before saving.

How it works in Halves

The NLP engine in Halves uses a combination of regex-based date extraction and Apple's NSDataDetector framework to identify structured information inside natural text. Here is what it recognizes:

Dates and times

The parser understands relative dates (today, tomorrow, next week, next Monday), specific dates (March 15, 3/15), and informal expressions (this weekend, end of month). It uses NSDataDetector for robust date parsing, which means it handles the same natural date formats that iOS itself understands.

Priority levels

Include "high priority," "low priority," or "medium priority" anywhere in the text and it gets extracted. You can also use shortcuts like "high pri" or just "urgent" for high priority. The priority keyword is removed from the task title so you get a clean name.

Assignment

Add "for me" or "for partner" (using their actual name) at the end of the input, and the task is automatically assigned. In a couple's app, there are only two possible assignees, so the parser just needs to determine which one you mean.

The chip preview

After you type your input, Halves shows a preview with colored chips for each parsed element: a date chip, a priority chip, and an assignee chip. If something was parsed incorrectly, you can tap the chip to adjust it. If everything looks right, tap once to create the task.

This preview step is important because NLP is not perfect. Sometimes "Friday" in your text is part of the title ("TGI Friday's reservation") rather than a due date. The chips let you catch and correct these edge cases without slowing down the common case.

How this compares to Todoist

Todoist pioneered natural language task creation and their implementation is excellent. It handles recurring patterns ("every weekday at 9am"), project assignment ("#Work"), labels ("@important"), and more complex scheduling than Halves currently supports.

Where Halves differs is in the couple-specific parsing. Because Halves knows there are exactly two users, it can parse assignment naturally ("for partner") without needing special syntax. And because the app is focused on shared household tasks rather than complex project management, the NLP can be tuned for the kinds of inputs couples actually type: errands, chores, appointments, and reminders.

Todoist also requires special characters for certain parsing (# for projects, @ for labels, p1-p4 for priority). Halves uses plain English for everything because there are fewer concepts to disambiguate.

Why speed matters

The difference between a 2-second task creation and a 15-second task creation is not just about convenience. It determines whether people use the app at all.

When you think "oh, we need paper towels," you have a two-second window before that thought gets displaced by whatever you are doing. If adding it to your shared list takes two seconds, you will do it. If it takes fifteen seconds of navigating menus, you will probably just try to remember it. And then you will forget.

Fast task creation is not a nice-to-have. It is the difference between a shared list that stays current and one that is always out of date.