Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

TerraPulse 🌿 — The Screenless Outdoor & Trail Agent

Description and use

TerraPulse cuts screen time to under 30 seconds, delivering concise audio briefings and micro‑quests for outdoor engagement.

Problem

Outdoor apps keep users glued to screens for trail info, reducing time spent outdoors.

Outcome

TerraPulse reduces screen time to under 30 s, providing actionable trail guidance and micro‑quests while enabling local inference and fallback options for unreliable connectivity.

Workflow

  • Select mission (Hiking, Trail Run, Foraging, Gardening)
  • Explore environmental inputs (temperature, humidity, soil moisture, elevation, canopy density, recent rainfall)
  • Generate Grass Score and trail‑condition assessment via TabPFN
  • Translate predictions into field briefing and micro‑quest with Gemma 2
  • Narrate briefing with ElevenLabs Turbo v2.5 or browser Web Speech API
  • Listen, put phone away, and engage with nature

Evidence

TerraPulse is an AI‑powered field guide for hikers, trail runners, foragers, and gardeners.

Source

I integrated Prior Labs' TabPFN (tabpfn.TabPFNClassifier) to analyze tabular environmental observations.

Source

TerraPulse integrates Google's open‑weight Gemma 2 (gemma-2-9b-it) as its naturalist‑style reasoning agent.

Source

TerraPulse supports ElevenLabs Turbo v2.5 for spoken narration, helping users listen rather than continually check their screens.

Source

The project uses sentry-sdk to instrument parts of the agent pipeline with custom spans.

Source

  • Record date: 2026-10-11T06:41:06.555Z