Step Out

Developer Cynthia Wahome built a local AI outdoor break planner called Step Out, using the Gemma 3 1B model with Ollama to generate outdoor activity plans on a local computer, without needing a cloud inference API.

Step Out lets users choose four preferences (available time, outdoor space, goal, action preference), then returns an activity card containing instructions, a time schedule, and reflection prompts, with no account registration required.

In Step Out's technical architecture, the Next.js backend sends generation requests to the local 127.0.0.1:11434 port, and reflection content is saved in the browser's local storage.

The Step Out backend handles time calculations (allocating 1 minute for preparation, 1 or 3 minutes for return, and the rest for the activity), and the model is not responsible for deciding the time allocation of the break.

Step Out uses a JSON schema generated by Zod 4 to constrain model responses; selectable activity IDs form an enum in the response schema, passed through Ollama's format parameter, and the backend then validates the returned JSON and checks whether the activity is in the selectable list.

Step Out uses Next.js 16's App Router and TypeScript, placing the interface and API routes in the same project.

On October 9, 2026, the developer tested a ten-minute seated activity in the yard, chose "quiet pause," and the generated activity was "notice outdoor sounds," which actually took 9 minutes and 30 seconds.

Step Out's demo version handles only one generation request at a time, and the current version does not support running on mobile phones.

Before inference, the Step Out backend filters from a curated set of five activities based on the chosen space, goal, and action preference; a seated choice excludes walking-type activities.

The Step Out project was completed in the first week of Hacktoberfest 2026.