Raphael KimCreative Technologist Let’s talk
ALL EXPERIMENTS02 / 07 — Taste becomes part of the system

Pucclab

A physical lab for the next espresso

How can an object help you learn from the taste of the last shot?

MY ROLE
Concept, interaction, ESP32 firmware, backend, notebook interface and AI recommendation loop
CONTEXT
Independent exploration of connected objects and sensory feedback
STATUS
Working prototype
Pucclab's white physical controller, e-paper display and yellow dial.
INTERACTIVE OBJECT / 002PROTOTYPE STUDY
DRAG TO TURNAN INTERPRETATION OF THE PROTOTYPE

Begin with a small decision.

A physical dial and display put the next coffee experiment within reach, right beside the machine.

PROTOTYPE / IN MOTIONThe complete interaction, from choosing a coffee to tasting and planning the next shot.

From input to experience.

THE INTERACTION LOOP
  1. 01

    Prepare

    Choose a coffee and set the recipe on the controller.

  2. 02

    Brew

    Use the physical paddle and follow the live extraction.

  3. 03

    Taste

    Record the sensory result alongside the measured shot.

  4. 04

    Explore

    Repeat a good result or try an informed adjustment.

01 / The ritual

Stay with the coffee

Dialing in espresso means repeating small adjustments to dose, grind, temperature and brew ratio. Pucclab brings the recipe, measured extraction and tasting result into a single physical workflow.

Select or scan a coffee, adjust the recipe and brew with the machine's paddle. During extraction, the handheld display becomes a live gauge for flow, time and weight. The result then leads directly into a seven-axis tasting.

The Pucclab controller held in two hands, showing a suggested next espresso test.
The same handheld object carries the experience from recipe to next experiment.
02 / The feedback

Advice grounded in the shot

The recommendation loop considers the recipe, extraction, tasting scores, coffee details, roast age, equipment and recent shots for the same coffee. It first decides whether the next step should correct a problem or explore a different sensory direction.

A good result can be repeated unchanged. When exploration makes sense, Pucclab can suggest more body or more clarity and explain what to look for. Grind changes stay on the grinder's real integer steps.

Pucclab shot history comparing recipes, extraction results and grinder settings.
Measured shots remain comparable after the coffee is gone.
03 / The instrument

Connected, with local control

I built the interaction, ESP32 firmware, backend, notebook and recommendation loop. The handheld combines a monochrome memory display, rotary encoder, scanner, battery and progress lights; ShotStopper inside the machine handles brew-by-weight control and telemetry.

Timing-critical control stays local to ShotStopper. The physical paddle and manual bypass remain available if the backend is offline, while the handheld caches its coffee catalog for browsing during reconnection.

The Pucclab Coffee Notebook showing coffee cards, roast information and shot counts.
A local notebook gives each coffee, recipe and tasting a persistent place.
WHAT THIS PROJECT EXPLORES

Taste closes the loop between a connected object, measured data and the next physical action.

Physical computing / ESP32 / Interaction design / Applied AI
Back to selected work Explore the full archive