Connecting to sources…

Peeqo (GIF-speaking desktop companion robot) · Voice-activated desktop companion that answers almost entirely in GIFs and short videos: a screen in the head shows eyes and GIFs, a three-servo Stewart-style platform nods and tilts the head, with an LED ring, speaker, microphone and buttons, a Raspberry Pi inside and a cloth cover

Voice-activated desktop companion that answers almost entirely in GIFs and short videos: a screen in the head shows eyes and GIFs, a three-servo Stewart-style …

Maker
Abhishek Singh (GitHub: shekit)
Form
Voice-activated desktop companion that answers almost entirely in GIFs and short videos: a screen in the head shows eyes and GIFs, a three-servo Stewart-style platform nods and tilts the head, with an LED ring, speaker, microphone and buttons, a Raspberry Pi inside and a cloth cover
For
Makers and developers: the official site calls it a DIY kit you build yourself (now sold out), and the README says you can create custom commands with the wiki tutorial even without a Peeqo
Sensors
Microphone and speaker: the dev kit uses a WM8960 sound chip (next-generation software docs), while the 2017 design has a Knowles mic board (Eagle files) and its prints include a cover for a Sabrent USB sound card; a 12-LED APA102 ring, 4 buttons and a screen in the head. Camera: the 2019 command list includes turning the camera on and off (a Raspberry Pi camera, python/zero.py) and the 2017 prints include a camera mount; the next-generation software drives no camera
Connectivity and app
Raspberry Pi Wi‑Fi. Next-generation software: with no known network it opens a “Peeqo-Setup-XXXX” hotspot and shows a QR code on its face for phone setup; a phone or computer on the same network opens a PIN-protected web panel at http://peeqo.local to talk, configure and calibrate. The live brain calls models through OpenRouter (Claude and Gemini only, defaulting to Gemini Flash) and fetches GIFs from Klipy; the offline brain only has written answers and cannot understand speech. The 2019 version used Google Dialogflow for speech, Giphy and Vlipsy for GIFs and videos, and OpenWeather for weather
Subscription
Self-built open source with no subscription; OpenRouter and Klipy run on your own API keys (the docs say the Klipy key allows about 100 calls an hour and the software budgets 90)
Openness
2017 version: peeqo-robot-hardware (STL prints, an IGES full design, Eagle circuits for an Arduino breakout, mic board, NeoPixel ring and servo board) and the peeqo-robot software are both GPL‑3.0 (repo-level LICENSE). shekit/peeqo, home of the 2019 and the 2026 next-generation software, has no licence file (no licence declared; the npm field in electron/package.json says ISC and the root package.json names none). The 2019 dev kit's mechanical and circuit files are not in the repos; the next-generation simulator carries only a 3D model of the dev kit, peeqo-devkit.glb
Privacy
A microphone is on board, and a 2019 command turns on a camera. Next-generation software: the wake word is detected locally by microWakeWord, and conversation audio goes to the OpenRouter model as a recording with no transcription step; facts the model keeps are stored on the robot in memory.md and sent through OpenRouter with each message; API keys, the hashed PIN and calibration sit in ~/.peeqo/config.json, readable only by Peeqo's user. The 2019 version detected its wake word offline with snowboy and used Google Dialogflow for speech recognition.
Not stated
Current price (the dev kit is sold out), parts list and total, size and weight, recommended age, the creator's location, hardware files for the 2019 dev kit

Vendor claims

  • A voice-activated desktop companion that responds entirely through GIFs (official site)
  • A three-servo Stewart-style platform moves the head, with expressive eyes and an LED ring: the next-generation software has 16 gestures and 23 expressions
  • The 2017 prints, circuits and code are GPL‑3.0; in 2026 the creator rewrote the software in TypeScript on the next branch

Parts

  • WM8960 · Handles microphone capture and speaker audio playback.
  • APA102 · Drives twelve LEDs forming the face ring.
  • PCA9685 · Drives the three servo arms over I²C.
Parts and sources

Sources