A product teardown, written in January 2025 when ChatGPT shipped its Tasks feature.
ChatGPT added task management. That should worry Todoist less than the coverage suggested, and more than Todoist's own positioning implied.
Here's the case for both halves.
Where Todoist is genuinely safe
Conversational task capture is a shallow moat. Anyone can build "remind me to call the plumber," and now several people have.
What's harder is everything that happens after a task list stops being a list. Dependencies between tasks, team collaboration, projects that run for months, and the tedious plumbing that keeps a task in sync with a calendar and an inbox and whatever else the user already lives in. Todoist has spent years on that layer, and a chat interface bolted onto a language model is nowhere near it.
Casual users will drift to whatever is already open in another tab. That was always going to happen and losing them costs less than it looks like, because casual users of a task app were never the revenue.
Where the position is weaker than it looks
Todoist is very good at recording what you said you'd do. It's close to useless at what happens when you don't.
Miss a task and it rolls over silently, forever, and the list slowly becomes a monument to intent. There's no cost, no acknowledgement, nothing that makes the user confront the gap between what they planned and what they did. A carry-over list, or anything that names an overdue task as overdue rather than just re-displaying it, would do more for retention than another integration.
The second gap is switching cost, in the wrong direction. Importing subtasks from other task managers is awkward enough that people stay where they are, which protects Todoist from churn out and equally blocks churn in. If you believe your product is better, friction at the border is not your friend.
The third is harder to name and probably the most interesting. Sometimes people don't want a list. They want to be told what to do next, in order, without having to decide. Not therapy, not life coaching, just a mode that removes the decision. Anyone who has watched Fleabag will recognise the impulse. A tool that already holds your deadlines and your history has more to work with than a chat window starting cold each time.
Three things worth building
Speech-to-text capture is the obvious one, and the least interesting, but mobile task entry by typing is a genuinely bad experience and fixing it is cheap.
Gamification is the one product people tend to sneer at. Streaks, badges, progress, rewards. It works, demonstrably, in categories where the user's real problem is motivation rather than organisation, and task management is largely a motivation category pretending to be an organisation category.
A productivity dashboard closes the loop that the missing accountability opens. Show someone what they actually completed against what they planned, over weeks rather than days, and the app has something to show them beyond a list of things they meant to do.
The real competition
The threat is not that ChatGPT does tasks better. It's that a general assistant sits closer to where the user already is, and proximity beats capability more often than product people would like.
Becoming more conversational is not the answer to that. Being the tool that notices what you skipped, and says so, might be.