About
My work is grounded in two convictions: event structure improves the reliability of model outputs, and time is a foundational primitive for memory, planning, and action.
I'm interested in memory interfaces and human-computer interaction: tools that preserve context over time for planning and reflection.
Runtime Labs
Runtime Labs is an independent research and product company focused on time-grounded language model interfaces. We develop systems that anchor language models in time, location, and events so natural language stays grounded in real-world context.
Óra
Óra is built as a planning environment where models operate on persistent, time-structured state. Plans are preserved as timestamped events anchored in time and location, with relevant notes, photos, and links attached to each event.
Schedule with natural language and images. Óra interprets intent, proposes events for preview and approval, and updates your timeline once you commit them.
Background
I did not anticipate starting a company. My work began with an interest in how brains process auditory signals and translate sensory information into natural language understanding. In time, I gravitated toward designing interfaces for time-based human-computer interaction, stumbling from early research prototypes toward what became Óra.
Current Work
| Planning interfaces | Natural-language planning across timelines, calendars, and notes |
|---|---|
| Time-indexed memory | Preserving model interactions in relation to past and future events |
| Human-agent collaboration | Shared surfaces where people and models can plan, revise, and commit actions |
| Mobile and voice timeline | iOS and web interfaces for persistent event context |