How I Finally Built a Task System With Todoist That Actually Stuck

I've used Todoist every day for three years, and for the first eighteen months of that it was basically a guilt list — thirty half-finished projects, no consistent way to add a task, and a home screen that reminded me what I hadn't done instead of telling me what to do next. This is the system that actually changed that. Not a redesign. Six deliberate changes, in order.

This is beginner-to-intermediate — you don't need Todoist's paid plan for most of it, though two steps below (Task Assist and Filter Assist) are Pro-only AI features. If you're still weighing which productivity app to actually commit to before you build a system in any of them, the three productivity apps I kept after trying fifteen is the more honest starting point; this guide assumes you've already picked Todoist and just want the setup that makes it stick.

The problem this actually solves

Task apps don't usually fail because the app is bad. They fail because the system inside it never got built — you download it, add forty tasks in an excited first hour, and by week three you're back to a sticky note because opening the app feels like admin rather than help. I did that same cycle at least four times with Todoist itself before any of it stuck.

What actually changed wasn't a better app. It was cutting the number of projects, giving myself one guilt-free way to capture a task the moment I think of it, and building exactly one view I open every morning instead of scrolling through six. The AI features Todoist added this year — Task Assist and Filter Assist specifically — turned out to remove the two steps I was most likely to skip: breaking a big task down, and writing a filter query by hand.

Step 1: Start with three projects, not thirty

My first Todoist setup had a project for almost every area of my life — eleven of them within a month. It looked organised and it was unusable, because every new task meant a small decision about which project it belonged in, and that friction is exactly what makes people stop adding tasks at all. Rebuild it broad instead: I run three — Work, Personal, and Someday — with sections inside each one (Work splits into Meetings, Deliverables and Admin) so related tasks group without the project list itself fragmenting. You can always split a project later once it's genuinely too big for one list; you can't easily tell in week one which of eleven projects deserved to exist.

Step 2: Capture first, organise later

The single habit that matters most is adding a task the second you think of it, on whichever device is nearest, with zero friction. Todoist's quick-add box understands plain English dates and times as you type — "call the accountant every first Monday at 9am" becomes a correctly recurring task without you touching a date picker — so there's no reason to delay capture until you're "properly" organising later. Later never comes; the task you meant to add tomorrow is the one that gets forgotten. If you also want an AI calendar that plays well with this kind of quick capture, I compared the options in the AI scheduling apps actually worth using.

Step 3: Label by context, not by project

Projects answer "what is this," but the question I actually ask myself during the day is "what can I do right now" — and that's a context question, not a project question. I use a small set of labels that cut across every project: @computer, @errands, @calls and @low-energy. When I've got twenty minutes at a desk with focus, I filter by @computer and ignore everything else, regardless of which project each task technically lives in. This is the step people skip because it feels like extra setup, and it's the one that turns a task list into something you can actually work from in the order your day allows, not the order you wrote things down.

Step 4: Let Task Assist turn a vague task into a real one

This is the first of Todoist's AI Assist tools, and it fixes the step I used to avoid most: breaking a big, vague task into the actual pieces that get it done. Type something like "launch the podcast episode" into a task, and Task Assist suggests a realistic set of subtasks — write the outline, record, edit, upload, promote — plus due dates based on your existing workload, which you edit or accept rather than write from scratch. It doesn't know your specific deadline pressure, so I still adjust the dates it suggests, but starting from a broken-down task beats staring at one vague line that's easy to skip past for another week.

Step 5: Build one Filter Assist view you actually open every morning

Todoist's filter query language is genuinely powerful and genuinely fiddly to write by hand — which is exactly the kind of setup step that gets postponed indefinitely. Filter Assist removes that: describe what you want in plain English, something like "overdue or due today, sorted by priority, not labelled @someday," and it writes the working filter for you. Pin the result as your home view. The point isn't having more filters — it's having exactly one you trust enough to open first thing, instead of scrolling past six projects to work out what actually matters today.

Step 6: Run a ten-minute Friday review

Every task system I've abandoned died the same way: not in a dramatic collapse, but through small neglect — a due date slipping unnoticed, a @someday task nobody ever revisits, a project quietly becoming a graveyard. Ten minutes on a Friday afternoon fixes this: clear anything overdue that no longer matters, glance at @someday for anything worth promoting into an actual project, and check next week's calendar against what's due. This is the step that decides whether the other five survive past a month, and it's the one I'd protect over adding any new feature.

Common mistakes worth avoiding

1. Building thirty projects on day one. It feels productive and it's the single biggest reason task systems get abandoned by week three — the decision fatigue of filing every new task is worse than not having a system at all.
2. Skipping the weekly review because "the app is up to date." A task list can look current and still be quietly rotting underneath — overdue items nobody's looked at, a @someday list nobody revisits. The review is what catches that.
3. Trusting Task Assist's due dates without checking your own calendar. It's working from your task history, not from the meeting you haven't logged yet. Treat its dates as a strong first draft, not a final answer.
4. Chasing Todoist's karma streak over actually finishing what matters. A streak rewards checking boxes, not checking the right boxes — don't let a long streak of small tasks quietly replace the one important thing you've been avoiding.

If the honest problem underneath your task list isn't the app but everything else pulling your attention away from it, I ran a genuinely different experiment on that front — what actually changed when I turned off every notification except calls for two weeks — and the two fixes turned out to work best together, not instead of each other.

And if you're still not sure Todoist is even the right tool for the job, I put it through a proper month-long test against the free app already sitting on every iPhone — Todoist vs Apple Reminders: which one actually gets things done — and the answer turned out to be less one-sided than I expected going in.

The two features doing the real work here — Task Assist rewriting a vague task and Filter Assist building the view you open every morning — are the same pattern I keep coming back to in everything else I test: a tool that drafts, and a human who decides. The rest of what has survived that test is collected in my AI tools and guides hub, if you want the setups rather than the hype.

Questions people actually ask

Is Todoist free, or do I need Pro for this system?

The projects, labels, quick-add and weekly review in Steps 1–3 and 6 work fully on Todoist's free plan. Task Assist and Filter Assist (Steps 4–5) are Pro-only AI features, so try them during a free trial before deciding whether they're worth paying for — the rest of the system holds up without them.

Why Todoist over Things, TickTick or a plain notes app?

Todoist's edge for this system specifically is natural-language quick-add on every platform — phone, desktop, browser, email — plus AI tools that turn a vague task into a scheduled one without a second app. Things is excellent if you're Apple-only and don't need the AI layer; TickTick's habit tracker overlaps some of the same ground. The system itself — three projects, context labels, one filter, a weekly review — works in any of them; Todoist just made building it faster for me.

Do the AI features read my emails or private notes to do this?

Task Assist and Filter Assist only act on what you type or select inside Todoist itself — a task you're creating, or a filter you're describing in plain English. Email Assist is opt-in per message and only extracts action items from an email you deliberately forward or connect, not your whole inbox.

How is this different from building a "second brain" in Notion?

Different job. A second brain is built for reference material — notes, decisions, project docs you'll search later. Todoist is built for the narrower question of what to do next, and trying to make one tool do both is usually what makes a system collapse. Keep your reference library and your task list in separate places built for each.

What if I fall off the system after a few weeks, like every other app?

That's normal, and it's usually the weekly review that slipped first, not the whole system. Todoist's karma streak is trying to solve the same problem, but it only helps if you're motivated by visible progress — if you're not, put a recurring Friday task in your calendar for the review itself, so restarting the habit doesn't depend on remembering it.