Project Diaries: Part I
Background
I’ve been tinkering on spare time projects alone for years.
The big problem has always been staying disciplined, maintaining motivation and shipping. I’d get really excited about every idea, but as soon as I told a few friends/family, Steven Pressfield’s Resistance would take over and I’d fail to ship. As I’ve come to realise, without shipping and without genuine feedback, it’s very difficult to tell if your projects are any good or not.
6 months ago, I decided I’d try a slightly different approach..
Strength in Numbers
My initial idea was simple: partner up with someone who would keep me accountable and give me honest feedback.
I needed someone who I could rely on, someone who would fully commit, and, fortunately for me, I knew just the person. Unlike me, Liam, my old university roommate, had already seen some success in monetising some of his spare time projects and so I reached out to see whether he’d be interested in working together on something. We met over a pint and to my excitement, he agreed to work on something together. We barely spoke about project ideas; Liam wanted to nail down how we would operate, what time commitment we were making to one another and how we would get back on track after the setbacks we would encounter. The main thing we decided was that no matter what, we’d meet every Monday at 8pm to check our progress from the week before.
Business partner secured, the next challenge was to decide what were we going to build?
Humble Beginnings
Fail fast and iterate, is it that easy?
One thing that both of us aligned on quite quickly was that we wanted to get something in front of our potential users as quickly as possible. To do that, it made sense to utilise the different skillsets that we individually had: Liam specialises in electronics design and development where as I’m more passionate about software development. To get started, we focussed on addressing problems that we had recently encountered in our lives and quite quickly, Liam mentioned a problem that seemed to fit our criteria: he couldn’t get a smart meter for his gas and electricity usage installed because of the age of his property.
Was this a big enough problem to tackle? I was skeptical.
To my amazement, some initial market research indicated that it was.
As of March 2023:
- Only 57% of all meters in Great Britain were smart, over a decade after the rollout began.
- Around 3 million (9%) of smart meters were not working properly.
- Components for an estimated seven million smart meters (out of around 32.4 million installed) will need to be replaced because they will lose functionality when the 2G and 3G mobile communications networks are closed.
We concluded that the current solution to reading electricity and gas readings was not accessible to everyone (and may not be in the future) and had some technical challenges both with installation and technological obsolescence.
Initial Idea
With rising energy bills being a hot topic, we knew that any solution we came up with would have at least match the smart meter’s unique selling point of being able to automatically submit periodic meter readings to a user’s energy providers.
Eventually we settled on what our MVP would look like: a Wifi enabled camera module, capable of submitting images of gas and electricity meter readings to a cloud storage location. The user would fasten the camera module near to the meter so that it could capture the desired images of the readings. Upon receiving images from the camera module, a backend would use computer vision techniques to digitise the meter reading data from the camera images. A frontend mobile app would allow a user to retrieve their latest meter readings.
This fitted perfectly with our skillset as Liam could focus on the hardware design whilst I wrote the firmware for the camera module and the software for the mobile app and so… we got to work.
6-Month Check-In
Working during our nights and weekends, we’ve managed to develop our first end-to-end solution which Liam is currently doing our first round of physical testing on 🎉
Via the mobile app, the user is able to connect to the camera module and give it access to the user’s WiFi network. From there, the camera changes into capture mode where it wakes up to take a meter reading capture every 30 minutes (replicating the frequency at which typical smart meters take recordings). Back in the mobile app, the user is able to see a list of all previous meter reading captures in digital format that can be shared with their energy provider.
The project has challenged us to overcome some significant challenges that has tested us to enhance our C++ skills for the firmware, PCB design for the electronics, Expo and React Native for the frontend and Javascript for the backend.
Major Problems to Overcome
The two major blockers that we’ve recently faced are:
- The hardware battery life
- The business case (given the high camera cost vs a free smart meter).
To get going (both in terms of practicality and cost), Liam targeted off-the-shelf hardware solutions for the camera module. This allowed us to quickly start iterating on firmware implementations but one problem remained: ensuring that we could run this module remotely for a sufficient period of time. A poor battery life would undermine one our unique selling points of restricting how many times a user has to access their meters to take readings. If the user had to charge up the battery every month, then we would have just substituted out one problem for another — not desirable. To overcome this, Liam has designed a custom PCB that completely cuts all power periodically, unlike the off-the-shelf solution.
We were grateful to make progress on the first blocker but there still remained one elephant in the room: how could we compete with smart meters on price when they are issued for free?
The project has two cost sources:
- The cost of producing one unit of the camera module
- The cost of hosting the backend database.
No matter how we tried to define the pricing strategy, there was no way to justify the cost to the user given the competitiveness of the smart meter offering. We started getting there with the cost of the camera module once we started factoring in sizeable economies of scale which got me thinking: were there use cases other than for taking meter readings where users would want to digitise analogue data streams by using computer vision? Could we make a general enough solution to cater for different use cases?
My answer was… yes.
Buildspace and Next Steps
Despite all of our progress, we still stumbled into the same old pit of iteratively refining a project without taking it to market.
We decided that to take our project to the next level, we’d need some accountability to push us to actually shipping what we’d built. That’s where Buildspace came in. I’d known about Buildspace for a long time but I hadn’t been brave enough to apply. Once I explained to Liam what the concept was, however, he convinced me that we had nothing to lose! At the very least we’d have a few more eyes on our project to steer us in the right direction.
We applied a few weeks ago and found out last week that we’d been accepted!
The next step is to use the 6 week programme to iteratively improve our product offering using actual feedback from potential users.
We’ve made the initial concept work for taking meter readings but I think it would be cool to spend the next 6 weeks finding a potential user who has a dedicated use case and taking what we’ve learnt to apply a general solution to cater for their requirements. I have some initial ideas that involve natural language inputs to determine what data a user wants to extract from image and video inputs from the camera. We can then manipulate the data to be displayed on the frontend in the most effective way possible for a particular use case I’ll write more about my ideas in the next diary entry!
The project may not be a success, but at least we’ve turned up consistently each Monday at 8pm 🫡
Note: Please feel free to reach out in the comments or message me privately if you have any feedback or need any help with anything!
