We had first-hand experience of building a Hardware product (MyT, Chai Brewing systems for Offices & Cafes). This blog outlines in simple words some of the challenges we faced while developing MyT. In this blog, I won’t touch upon the business challenges that we faced but only the product challenges.
About MyT (pronounced as mytea) – it is a Chai Brewing System for Workplaces and F&B setups. MyT automates the Indian-style Tea (Chai) making process by boiling all the ingredients (CTC tea, real spices, milk, water) together. MyT neither uses Tea Bags, premixes & coffee process to prepare Chai, nor does MyT stores pre-prepared Chai (Read more on types of Chai Machines). It is a novel Brewer that employs the preparation process and ingredients that are used in Indian homes. If you would like to know more about MyT, get in touch!
Before I start, consider the following points to set the context:
- When we started MyT, we had zero experience of building a Hardware product
- We bootstrapped MyT and were extremely tight on money throughout our journey
- We wanted to build something that many large companies failed at in last couple of decades – a machine that can make Chai which India loves.
- We really really wanted to make it work
- We were based out of Bengaluru, India
Hardware product are challenging because they are money, time and effort guzzlers. Additionally, they require knowledge of multiple disciplines, undergoing a long R&D process, implementing market feedbacks and then cracking Mass manufacturing.
In the following paragraphs, I will touch upon aforementioned points and provide examples (where I can) from our journey about the challenges Hardware startups in India in the mentioned context.
Hardware Requires Multi-disciplinary Expertise
Hardware products require knowledge of multiple disciplines. I will try to be specific and mention briefly what all did we needed to know while building MyT.
Understanding of relevant Engineering
Generic beverage machine engineering , thermal systems, fluids & thermal engineering
Fundamanetals of Electronics/Electrical Engineering, Drivers, Microcontrollers, Interfacing of various hardwares
Different Motion mechanisms, Motors (DC, Stepper, Servo), Structural aspects
Generic programming skills and how to program various hardware components
Sheet metal manufacuring, Plastic manufacturing/fabrication, Rapid prototyping, Jugaad protyping, Machining, Moulding etc.
General Design thinking, Industrial design
So, if the founding team can handle these areas, great! If not, then the startup needs to acquire these skills themselves, hire or outsource. Acquiring skills can be most effective but time consuming & frustrating affair; hiring is fine if startup can pay salaries. Outsourcing is a tricky and comes with its own set of problems (quality, delays, expensive). What one does really depends on their inclination and their constraints.
R&D Process is long and expensive
R&D involves coming up with a workable product idea, prototyping it fast and cheap, testing it, learning from it and then improving upon it. This is repeated till one hits something which is worthwhile of further development. It sounds simple but in reality it requires relentless creativity, mountains of motivation and endless optimism to keep generating new ideas and ruthlessly rejecting ideas that don’t work.
Prototyping takes time
A typical prototyping involves setting up hardware (sensors, heaters, pumps, motors, valves etc), interfacing hardware to the circuit, programming the board to run as desired and if needed build an enclosure around it. If the function is not final then the idea is to see whether the prototype functions the way one wants it to. If it does the prototype is developed further, if not a new concept / idea is worked upon.
I’ll give an example. We prototyped thermoelectric cooler to cool the stored milk so that milk does not spoil at room temperature. The aim of making this prototype was to check how many peltiers it takes to cool milk down and find out the rate of cooling.
The prototype cooled 1.5 litres of mik by ~20C in 2 hours with an efficiency of ~3%. The conclusion was that we won’t go this way. Later, we learned that in a typical office more than 1.5 litres milk is consumed easily in less than an hour so milk does not reach a point where it gets spoiled. In the hindsight, we should have never tried this in the first place.
Building this jugaad setup was easy – fix cold side of peltier on a vessel, fix heat sinks with a fan on hot side of the peltier, connect it to power supply, program the sensor and run. But still it took slightly more than a month before we obtained the curve shown above. Things blowing up, adhesive coming off, different design, different thickness of vessel, improper insulation, sensor calibration, manufacturing of vessel and sometimes wrong ideas take away a lot of time. I wish something as easy as ctrl-C, ctrl-V, ctrl-Z, ctrl-X existed for physical products 🙂
Non-availability of components
As prototyping progresses, new ideas come up and the need for some specialized components/electronics arises. Usually such components are not available in India and need to be imported from EU, US or China. They components are expensive with very limited information available about their working; mostly they are available as spares of some existing products. It takes time for components to arrive, takes time to learn how to use them and sometimes they get stuck in customs which further delays things.
- Once we ordered an expensive heater from EU which was supposed to run in a specific way which wasn’t described anywhere; we ended up melting it during our tests and had to re-order.
- On another ocassion we wasted a month trying to run an arduino shield ordered from sparkfun US. After a couple of weeks of frustration, we ordered the second one which worked fine and we realized that we probably blew the first one on the first day of use.
- At another time, a seller from China decided to downvalue parts we had ordered and consequently those parts got stuck in customs. Later we had to let go of those and re-order as the penalty was way too much.
There are weeks where one would feel “what did I achieve in this week?”
Custom part development
Manufacturers don’t entertain custom part development in small quantity. Its a hassle for them as their machines are already occupied with the other ongoing job-works. Additionally, a new custom part development can’t be offloaded to a junior member and a senior person at the factory needs to get involved. It is time consuming deal for one-time money and most of the time there is no recurring business for the Manufacturer.
After a lot of searching around in Bangalore, we found a couple of manufacturers who were willing to help us on our prototypes and custom parts.
- One of them did development for us twice for four times the usual cost and later stopped responding to us.
- While the other reguarly took 2.5-3 months (on a 2 week delivery committment) for our parts with daily whatsapp / calls followups and weekly in-person followup visits (~100 km round trip).
- We wasted aggregate of 1+ years easily waiting for components to arrive, manufacturer to give us the parts and design modifications from the design firm.
- Lastly, it used be disturbing when you get a part after months only to realize that even after multiple discussions, the specifications are off and will need to be re-done.
Thankfully, last year we discovered IKP and got rid of time component of the aforementioned problems.
After churning and rejecting several prototypes, we were satisfied with our 25th. We were anxious / curious to give it out to customers on subscripton basis. But generally the sentiment at the prototype level was to crack functionality literaly at any cost. Which meant that some more work was required before we could actually have a version that can be pitched for a paid use. Specifically we now had to do the following:
- A decent enclosure was required that was required to be done without using plastics or incurring any fixed costs
- Some components needed to be optimized for cost and reliability
- We realized a couple of modules required re-engineering to ensure reliable and predictable performance so that we don’t have to fix them often.
After so many prototypes, we were shattered that another 6 months of effort was still there before a single rupee of revenue is earned; sadly there was no escape. Finally, the design firm delivered the CAD design in 1.5 months, manufacturer gave us 5 units in 3 months; in the meantime, we had changed engineering to be more reliable and were ready to install in the 6th month.
Product-Market Fit Evaluation
Once the paying customers are in, the real feedback pours in. We, particularly, were looking for the feedback on usablity, usefulness and functionality of the product.
- Usability was realtively quick and easy and within a month and talking to a few users told us what needed to be done.
- Usefulness was pretty straightforward but it took time – if the monthly consumption was stable for 6-month period then it implied that the users were repeatedly using MyT and find it useful enough.
- However, the functionality evaluation was difficult and time-consuming because of the following:
- Incorporating feedback was difficult if any hardware modification was involved. In general, this required going to installed locations, opening up the Brewer in office / cafe during off timings, installing the new part and then testing it then and there.
- Another aspect was evaluation of robusness of the components. Components / modules were observed for several months to see which component wears out followed by fixing it / replacing it and observing it again if it didn’t work. This required at least 6 months of operation in different settings so that it could amount to thousands / lacs of cycles of a component.
Took us almost an year and half to understand and be sure of what really worked in the real setting, what didn’t, what users wanted, what they didn’t, what we wanted and what we didn’t. Thereafter we ideated for a few months (~6-7) and fortunately we hit on a concept which we firmly believe would prove to be the dominant design for future Chai machines. Although the new concept required complete architechture overhaul of the 26th version, we were able to achieve that post 9-10 functional jugaad prototypes. This new version (36th) solves all our product problems that we experienced first-hand in the market. Particularly, the improvements over the 26th version were – improved taste in half brewing time, enhanced usability & automation, simple to manufacture, assmble & service, easier maintenance, and more robust.
This is the next item that we need to figure out in near future. I won’t write about it as of now but would update it later.