For the past few months, our four-person team has been quietly working on a problem that every productivity tool claims to solve, yet almost all of them make worse.
Today, we are launching the new version of InfiniteUp across iOS, Android, and the Web (tasks.infiniteup.dev).
This isn’t just a version update or a UI refresh. It’s the result of data gathered from thousands of global users, months of agency development, a self-inflicted bottleneck, and a technical breakthrough that changed how we build software. Here’s the story.
1. The DOER Legacy: Learning from 5,890 Users Across 117 Countries
To understand InfiniteUp, you have to understand why we retired its predecessor, DOER Business.
DOER Business was a success by most startup metrics — 5,890 users across 117 countries, zero ad spend, a 68.6% engagement rate. People genuinely loved it.
We retired it anyway. Usage data revealed a brutal truth: people weren’t short of places to write things down. They were getting stuck at the exact same moment every time — the item too big to start. A massive goal like “Sort out corporate taxes” or “Plan product launch” would sit untouched for months while smaller tasks got ticked off around it.
An overwhelming item isn’t a single task — it’s five tasks in a trench coat, and none of them have been named. A prettier checklist wasn’t going to fix that. The tool had to break the mountain down itself.
2. Designing “Invisible AI”
Moving from DOER Business to InfiniteUp meant months of architectural dialogue between our CPO, Wasim, and CEO, Barrett Nash. Looking at how the industry was bolting chatbots onto existing SaaS interfaces, we reached a clear conclusion: Main Street doesn’t want to manage an AI. It doesn’t want a side panel, a prompt, or a conversation to babysit.
So we committed to a different philosophy:
- Traditional AI: user writes a prompt → babysits a chatbot → manually copies the output.
- Invisible AI: user dumps a raw thought → AI structures it in the background → a deterministic UI shows the plan.
In practice, that looks like:
- Fast Add — dump a raw thought (“Prepare Q3 report for the board before Friday”) and keep moving. The system parses priority, sets realistic dates, and structures subtasks behind the scenes.
- Deterministic Day View — your daily view runs on rigid, non-probabilistic logic. What’s due today is a fact, not an AI guess.
- Active Task Partner — open a task and an AI agent is already there, editing it live and asking precise, multiple-choice questions to refine the plan in three taps.
3. The Agency Reality & the Dogfooding Trap
While designing InfiniteUp, our team was running its day job: a profitable software agency out of Nanaimo, British Columbia, with teammates in Tbilisi, Beirut, and Nairobi, shipping 40+ client apps that have reached over 60,000 installs.
Building a product alongside client work meant engineering time was precious. We had an early version of InfiniteUp we liked in theory but weren’t yet using ourselves. So we mandated that every internal team task run through InfiniteUp In InfiniteUp — over 1,000 test tasks created.
That’s when the friction showed up: UX bugs, edge cases, date-logic glitches, latency issues. Client work kept demanding our focus, and the backlog grew faster than our fixes — 126 clearly defined tasks and bugs piled up on our internal ClickUp board.
4. The Breakthrough: The “MCP Scalpel”
The turning point came when Nash deployed Claude Code alongside Model Context Protocol (MCP) servers to work directly inside our FlutterFlow architecture — a methodology detailed in our essay, MCP Is a Scalpel for App Development, Not a Chainsaw.
Using MCP, Nash interfaced directly with the project codebase and ripped through all 126 ClickUp tasks in a matter of days.
But the real lesson wasn’t AI speed. Claude Code couldn’t have fixed those 126 issues without months of meticulously documented user feedback, mapped edge cases, and internal dogfooding. The AI was the scalpel — the months of structured context were the hands guiding it.
5. What InfiniteUp Actually Does
Type It, Say It, or Photograph It
- Type a single line — Fast Add handles 82% of all tasks created in the app.
- Speak your mind — hold the voice button and talk naturally; the system infers context, priority, and deadlines from tone and phrasing.
- Photograph your desk — point the camera at a handwritten list or whiteboard and it’s parsed into structured tasks instantly.
The Task Partner
Every task includes a dedicated AI partner that modifies the task interface in real time. Tap “Make the first step smaller,” and the AI rewrites the subtask structure live on your screen.
Desktop & Mobile Alignment
On mobile, InfiniteUp is a streamlined, thumb-friendly daily feed with real-time voice and photo capture. On desktop (tasks.infiniteup.dev), it becomes a full Kanban board and drag-and-drop weekly calendar.
Frictionless, Task-Level Collaboration
Most project tools force you to create a whole “workspace” just to collaborate with one person. In InfiniteUp InfiniteUp, collaboration happens at the task level — create a task, invite anyone by email, and it lands on their board only when they accept.
6. First Users of Our Own Revolution
Today, the entire InfiniteUp team — across Canada, Georgia, Lebanon, and Syria — uses InfiniteUp as our primary daily driver, having migrated off Google Calendar, sticky notes, and third-party tools. We’re still polishing edge cases weekly, but the productivity gains inside our team are real and immediate.
InfiniteUp proves that AI in software doesn’t mean adding more chat windows to manage. It means interfaces that understand intent, eliminate cognitive friction, and turn what you’ve been avoiding into step-by-step progress.
InfiniteUp is completely free while we keep shaping it with our community. Bring the one task you’ve been avoiding for months, point it at the mountain, and let it hand you the steps:
