OmniFocus 3
Context
OmniFocus is a long-standing task management application with a deeply committed user base. By the time work began on version 3, the product had grown powerful, flexible, and complex. That complexity was part of its appeal, especially for experienced users, but it also created friction for people who did not want or need that level of control upfront.
I worked closely with the CEO and product manager to define the scope of the release while leading the UX team responsible for the redesign across macOS and iOS. My role included research direction, design strategy, and creative oversight, along with managing a team of five designers and researchers. I also partnered closely with QA and Support to incorporate real customer feedback into product decisions.
The work was not about adding features. It was about refining a mature system without destabilizing the trust it had built over years.
This case study predates the current wave of AI tooling. The “AI in Context” section at the end explores how AI could assist with task organization, prioritization, and system shaping, reducing setup friction without compromising flexibility.
The Core Tension
OmniFocus was originally built as a digital extension of David Allen’s GTD system. Long-time users understood its terminology and structure deeply. Over time, research and support feedback revealed growing friction. Core concepts—especially “contexts”—were difficult for new users to understand. Many customers were already describing them as tags.
Growing the product required addressing that gap. Changing a foundational concept in a mature product carried cultural risk. Power users trusted the system as it was. Altering it could feel like betrayal.
This was not surface polish. It required revisiting the product’s mental model.
Tension: modernize the system’s language without eroding the trust of long-time users.
Replacing Contexts with Tags
Replacing contexts with tags appeared simple from a feature perspective. It was not simple culturally.
Internally, contexts had existed since the product’s inception. For some team members and dedicated GTD users, they were foundational. Externally, however, expectations were shifting toward more flexible tagging behavior.
The design work focused on preserving power while improving clarity:
- A visual representation that scaled across outline and fluid views
- Capsule-style tag displays to preserve readability
- Overflow handling to prevent vertical sprawl
- Consistent iconography in sidebars and inspectors
The objective was not simplification for its own sake. It was reducing friction without reducing capability.
Post-launch monitoring showed strong adoption and significantly reduced confusion around contexts. The feared backlash did not materialize at scale.
Custom Perspectives
Perspectives were a defining capability of OmniFocus. Power users relied on them heavily to shape their workflows.
Research indicated demand for greater filtering flexibility, but expanding filtering risked overwhelming less technical users. Instead of building a fully bespoke system, we leveraged the macOS predicate editor framework. That decision constrained visual freedom but reduced long-term maintenance and platform fragility.
Users gained the ability to build precise “if this, then that” rules while staying grounded in native macOS conventions.
The expansion increased capability without increasing conceptual noise.
Forecast Integration
The Forecast view was the most-used perspective in OmniFocus. It displayed tasks due each day and optionally surfaced calendar events in a separate section.
Users reported friction between tasks and calendar events occurring sequentially in real life but appearing separated in the interface. Integrating calendar events directly into the chronological outline reduced that disconnect. Color-coded indicators preserved distinction without dividing the timeline.
The change reduced mental context-switching and improved clarity around how the day unfolded.
Customizable Inspectors on iOS
The inspector contained twelve potential controls affecting an action. For some users, that flexibility was essential. For others, it created cognitive load.
We introduced customizable inspectors that allowed users to rearrange and hide fields. My initial proposal extended further but would have required disproportionate engineering investment. After discussion, we implemented a measured version that delivered personalization without destabilizing the architecture.
Restraint protected long-term stability.
Platform Constraints
Throughout the redesign, there was tension between custom design and platform convention.
Earlier experience had shown that diverging too far from Apple’s SDK patterns creates long-term maintenance risk. For OmniFocus 3, we intentionally worked within system constraints. Custom components were introduced only when necessary.
Durability guided those decisions. The product needed to remain stable as Apple evolved its frameworks.
Outcomes
OmniFocus 3 launched to strong adoption and positive reception.
- Daily active users increased 12 percent on macOS and 17 percent on iOS
- The iOS version remained in the App Store’s top 25 grossing apps for six weeks
- Media reviews highlighted increased flexibility and control
Long-time users felt heard. New users found the system more approachable.
The release demonstrated that mature products can evolve structurally without abandoning the people who built them into what they are.
AI in Context
If I were approaching OmniFocus 3 today, I would be interested in how AI could support task organization, prioritization, and system shaping without flattening the flexibility that makes the product valuable to serious users.
A central challenge in OmniFocus 3 was reducing friction without reducing capability. The product had grown powerful, but that power came with structural complexity. Replacing contexts with tags, expanding perspectives, integrating Forecast more naturally, and making inspectors customizable were all decisions aimed at preserving depth while making the system easier to understand and use.
AI could extend that same direction. It could help users organize incoming tasks, suggest tags or projects based on patterns, identify likely priorities, or propose perspectives that match how a person already works. For less experienced users, it could reduce the burden of building a personal system from scratch. For advanced users, it could act more like a configurable assistant that helps maintain and adapt an existing system over time.
The risk would be over-automation. OmniFocus has always been valuable because it gives users meaningful control over their workflows. The right role for AI would not be to take over that system. It would be to help users shape it more quickly, adapt it more intelligently, and reduce the friction of configuration without compromising trust or flexibility.