Skip to main content

About the project

Pikseli Vision × ElevenLabs is a bilingual (Finnish/English) AI companion for makers, hacklabs and art lovers. Point your camera at a solder joint, a 3D print, a painting or a messy desk, and it looks at the image and talks back out loud in character.

Gemma 4 handles the vision. It reads the photo and returns structured output: a spoken response, an emotion, a verdict, and 3–4 detected details shown as tappable chips. Users can also ask follow-up questions about the image (“Why is this corner warping?”). Six personas, from Matti the Maker to RoastBot 3000 to Julian the Art Critic, change how it sees and what it says.

ElevenLabs (eleven_multilingual_v2) is the voice. Each persona has its own voice, speech starts automatically after analysis, and it falls back to browser speech if ElevenLabs is unavailable, so a maker with a soldering iron in hand never has to look at a screen.

Built at Hacklab Mikkeli, open source under Apache-2.0.

Built with

geminigemmaelevenlabs

AI tools

Coded withClaude CodeGemini 3.8 FlashClaude Sonnet 5.5Google AI Studio

Prize Categories Entered

  • Deepmind Google

    Best Use of Gemma 4

    Submitted
  • EElevenlabs

    Best Project Built with ElevenLabs

    Submitted