TextTheDo
Text your to-do list into existence.
A task manager with an iMessage-style interface: text it what you need to do in plain language and spaCy-powered NLP turns each message into a prioritized, categorized task that syncs to your calendar. It reaches ~90% accuracy on task recognition and is an active work in progress.
remind me to email the TA about the midterm before friday
Email the TA about the midterm
synced → Google · Apple Calendar ✓
The problem
Every to-do app asks you to stop and fill out a form — a title, a due date, a category, a priority. That friction is the whole problem. The exact moment you have to organize a task is the moment you decide not to bother, and the task never gets captured.
How it works
TextTheDo replaces the form with a conversation. You text tasks the way you’d text a friend — “remind me to email the TA before Friday” — and spaCy parses the message for intent, entities, and timing, then generates a prioritized, categorized to-do automatically.
Tasks sync in real time with Google and Apple Calendar, so what you capture in a text shows up where you already look.
Engineering decisions
I reached for spaCy rather than a large language model on purpose: task parsing needs to be fast, predictable, and cheap to run, and a focused NLP pipeline is easier to reason about and correct than a general model. The messaging metaphor is the product — every design choice serves the goal of removing the step between “thought” and “captured.”
Status & what’s hard
This one is honestly still in progress. The open problems are the interesting ones: pushing task-recognition accuracy high enough to trust without a confirmation tap, and reconciling how Google and Apple model calendars differently. I’m actively refining the NLP models and adding integrations.