Demo

Echo Demo Release

Echo is a Flutter safety assistant built for the Gemma 4 Good Hackathon. This release captures the current working demo: a safety flow powered by a local Gemma/llama.cpp server, local emergency state, optional ElevenLabs voice, optional Telegram contact escalation, and an Echo Feed for responsible community amplification.

What Echo Can Do Now

• Run a Chrome demo with voice SOS, typed fallback, and listening/thinking/speaking states.
• Use local Gemma reasoning through llama.cpp at http://localhost:8080.
• Warm up and verify Gemma through a live System Test drill before entering Home.
• Trigger emergency flow from the main SOS button or from high-risk voice/conversation output.
• Escalate through Tier 1, Tier 2, and Tier 3/Echo Feed states.
• Store returning demo user, emergency, and feed state locally.
• Post missing-person/community cases to the Echo Feed with exact map links hidden.
• Amplify Echo Feed posts for share-ready community response.
• Use ElevenLabs TTS through a local proxy, with local TTS fallback.
• Use an optional Telegram bridge for Tier 1/Tier 2 alerts and SAFE, HELP, or CALL replies.
• Fall back gracefully when speech, Gemma, ElevenLabs, Telegram, or network paths are unavailable.

Key Screens

• Splash / Onboarding: Introduces Echo’s safety mission and missing-person prevention framing.
• Auth / OTP: Local demo sign-up flow. Use OTP 000000.
• Profile Setup: Captures basic user details for demo identity mapping.
• Permission Setup: Explains microphone, location, and emergency-related permissions.
• Tier 1 Setup: Adds trusted inner-circle contacts.
• Tier 2 Setup: Configures wider escalation and Echo Feed readiness.
• System Test: Runs a live Gemma health check, context warm-up, and drill inference before Home.
• Home: Main SOS orb, voice conversation entry, Echo status, contacts, and safety actions.
• Gemma Intel: Demo console where Echo explains itself, its purpose, safety tasks, and future on-device direction.
• Emergency Active: Shows active incident state, escalation timer, Tier 1/Tier 2/Tier 3 progress, and contact response loop.
• Threat Analysis Result: Displays Gemma’s structured risk assessment and recommended safety path.
• Activity / Echo Feed: Shows local incident timeline, missing-person case posting, and public amplification cards.
• Settings: Privacy controls, cloud-processing notes, ElevenLabs/Telegram toggles, and model-download preview.
• Contacts / Profile / Notifications / Terms: Supporting account, safety, and transparency screens.

Demo Path To Try

• Sign up with demo OTP 000000.
• Complete profile, permissions, Tier 1, and Tier 2 setup.
• Run the Gemma System Test.
• From Home, hold the voice/SOS control and ask Echo who it is or request help.
• Trigger an emergency flow and watch Tier 1, Tier 2, and Echo Feed escalation.
• Open Activity to see the Echo Feed and missing-person case amplification.
• Open Gemma Intel to test Echo’s demo Q&A.
• Open Settings to see privacy controls and the model-download preview.

Transparent Constraints

• Chrome is the reliable, tested surface for this release.
• Direct Gemma WAV/audio understanding was tested experimentally with llama-mtmd-cli, but it is slow on the current CPU setup, so the app defaults to transcript-first Gemma reasoning.
• On-device Gemma inference is architecturally prepared but not claimed as fully validated without Android runtime/device testing.
• ElevenLabs and Telegram require internet and local secrets.
• The APK/model download flow is included as a preview path for future validation, not as the primary submission runtime.

Release Notes

This release is best understood as a functional web-first safety demo and repository snapshot. It demonstrates Echo’s core product loop: early signal capture, Gemma-assisted context, trusted-contact escalation, public Echo Feed amplification, and graceful degradation when optional services are unavailable.

First feature merge with Audio and Voice Service

What's Changed

Merge with feature and services by @ranmori

• UI Pull by @estheticallybawo in #1

New Contributors

• @estheticallybawo made their first contribution in #1
• @ranmori made the voice activation and audio recording service
• @maryprecious made the UI changes to meet our design requirements
• lilian for the Figma design

Full Changelog: https://github.com/estheticallybawo/echo/commits/v.0.1