After a non-negligible development effort, the initial proof of concept (POC) was finally mature enough to be faced with real world use cases.

I hooked up a bunch of different sensors and actuators to it, to test out what I could build with such a creative I/O box:

• a midi mechanical sequencer using some cardboard, an infinite servo, and a few IR barrier sensors. The sensors were reading the rotating sequence punched in the cardboard, sending notes to a DAW on my computer.  
• an Etch A Sketch game one can play like a theremin with distance sensors
• controlling a led strip with knob

These first attempts worked out decently, and using this new tool made things so much quicker to setup rather than starting from scratch with an Arduino that it comforted me on the added value of the project. But I also got highly frustrated by some of the (un)practicalities with the current design:

• Connectors, while having chosen them carefully for being snap-in and detachable turned out to be rather impractical.
• The wireless capabilities of the whole device was completely undermined by its size. This limited a lot the potential applications
• Being compatible with almost any sensor was not so great since a lot of them do not perform that well depending on situations or need specific processing to get practical signals from them

On the other side, this proof of concept also gave me a glimpse of the remaining very large development challenges I would have to face to bring such an initial concept to a rather mature state.
Aside from the technical and design issues, I also had to realize the too large application scope resulting from too many features with too much flexibility made the communication around the project very difficult.
It was the right time to rethink the whole concept. Putting all this work in a drawer was a painful step to take, but moving away from it and starting all over again was the only way to leverage all the learning of the past months to rebuild on much stronger foundations.

Of course the new approach came out from playing around with the POC. Highly frustrated by the impracticable cables going to the sensors, I though about making them wireless. And this simple idea made me move entirely toward a fully modular approach which turned out to be the best answer for so many more aspects of the project than just the wiring. Splitting the project into modules brought many key improvements:

• each module is specialized to a limited feature set. Sensors and hardware is curated, processing optimized.
• modules can be much smaller, wireless, and therefore can be easily embedded into objects or used as wearable.
• the project can scale over time by adding new modules/features to the family
• modules all share the same code base simplifying the development

And that's how the development of the current shape of Pipo modules (re)started.

As a conclusion of this process, I would keep the following toughs for the future:

• Try out your concept as fast and early as possible (and without actually designing/developing if you can)
• Less is more. Focus on core features/functions.
• Get feedback early
• Identify what is the true added value in your project