Hardware Products Built for Scale

Product Design, Product Management, Lo-fi prototyping, Hi-fi Prototyping, Supply Chain Management, Manufacturing
Coral Robots
CEO, Co-Founder
May 2017 - Present
How do you build a consumer robot from scratch?
When it comes to designing a high-tech hardware product, it’s important that you have a super diverse team. At Coral, we had 4 multi-disciplinary teams working together to build a new home robot. Our electrical engineers, mechanical engineers, software engineers, and designers did everything together as a team.

I led our teams to take insights that we got from users and incorporate them into our designs. From there, we started gathering all the variables we had to consider: cost, materials, time, design, and so many more. From there, we started building our final product and get things out on the production line.

Just kidding, that’s when the prototyping and testing phase began. And you can bet that there was a lot of it 😉

In the News:
The Best Robot at CES 2019 via Business Insider
All the Gadgets WIRED Editors Loved This Month (November 2018) via WIRED
This New Robot Vacuum Transforms into a Handheld Vacuum via Android Headlines
How two brothers launched first 2-in-1 robot vacuum in months via Campaign Creative

Start with a Concept

Any great product must start with an idea. When I worked with my team, I encouraged them to be as wild and crazy with their ideas. When you design a product, I truly believe that it’s impossible to have the greatest, most brilliant idea. But if you come up with a bogus, unlikely, or crazy concept, you end up with something even more valuable than any other idea.

You end up with a hypothesis.

When we re-framed our ideas or concepts as hypotheses, it gave us actionable and measuring steps to start designing new products and experiences.

Understanding the User Journey

One of the best ways to understand where you might have interesting opportunities or problem areas is to map is out. After I see things visually, I feel like I can better interpret or synthesize any design challenge.

I led a multi-disciplinary team of designers, engineers, and sales people to map out the real customer journey of robotic vacuum consumers.

We broke it down into 4 stages: the shopping experience, the initial setup, daily usage, and product maintenance. Throughout this entire process, we interviewed 6 existing customers, 2 power users, and 2 people who didn’t even know what a robot vacuum was.

Drawing Key Insights

Mapping out the user journey gave us immediate problem areas to start working on. I worked with our business team to identify core part of the consumer shopping experience where we could be more accessible, interesting, and exciting than competing brands. I worked with our product team to rethink the entire cleaning experience. We had to take a step back and ask:

"How do people actually clean?"

And we found something really interesting. There was a huge gap between the expectation for robot vacuums and the reality of the user experience. We knew that we had to design a product that addressed this issue.

Cleaning Happens in 2 Parts

We had to close the gap between the expectations and reality of how people clean with technology.

So, I sat down with our product designers to create another set of surveys and discussion guides that could help us draw some new insights. In our second round of user research, we interviewed some full-time homemakers, professional cleaning services, and lazy college students who neglected hygiene.

And we learned that cleaning actually happens in 2 parts: cleaning for maintenance and cleaning specific messes. It helped explain the product diversity in our industry. Handheld vacuums existed for small, discrete messes like when snacks make it on the floor. Larger vacuums existed to do the maintenance unnoticeable filth that builds over time.

That led us to a key question: what if you built one product that does both?

Rinse and Repeat

This was by far one of the most challenging parts of building a hardware product. Before you can start shipping it, you have to try one idea, watch it fail, and then try something else.
I worked with our design team to create an aesthetically-pleasing form factor that could be technologically viable for a unit. We had to consider user expectations for form factor, size, shape, color, and more.

I worked with our hardware teams to source the parts that could fit our designs. We had to consider specs such as airflow, suction power, filter surface area, which were tested by testing our prototypes in wind tunnels, noise-cancellation rooms, and more. I recall one week where we ended up combining the design and hardware teams’ desks so they could optimize their iterations.

I also worked with our software team to design a low-weight architecture that could run anything from algorithms to AI maps. In order to keep our hardware production costs low, we had to write some of our own low-level firmware to get access to readily available parts. We could eventually map out an entire room, send information to the cloud, and process some minor graphics.

At the end of the entire process, we built over 5 1:1 scale foam core models, 30 3D-printed models, and 2 CNC-ed fully functional mockups.

Getting Ready to Scale

This is where things got serious. But I was fortunate to also stretch my product management muscles during this stage of our process. As we started nearing completion of our high-fidelity prototypes, we had to make sure that we could actually build our product at scale. We had to simulate best-case, expected, and worst-case scenarios and extrapolate timelines to see how we wanted to proceed.

I sat down with our engineering team and our manufacturer to design our process for assembly, production, QA, certifications, and shipping.

Manufacturing is a Delicate Dance

This was perhaps one of the biggest moments of learning for me. You don’t really realize how complicated hardware tech manufacturing can be until you go into the belly of the beast.

I recall immediately being hit with a barrage of questions: Which vendor is responsible for which component? Will those components be readily available or are they LLT (long lead time) parts? Are we using custom or common-use parts? Who is our 2nd source vendor? Do we know the vendor? The list can run on forever…

But we still somehow got through all of them and went into production.

Dot your i's, Cross your t's.

Check every last detail. Not only is this important in building hardware products but I think it’s a great life philosophy.

Before we hit the assembly lines, I held team briefs to talk about our each part of the production process. Some teams were in charge of QA, others in charge of inventory verifications, and I oversaw the main production flow.

In order to deliver the best experience possible to our customers, we had to pay attention to every little detail. I recall one time when I had to hand-check if gloves had holes. If there were holes, it meant that people with longer nails were more prone to scratching the plastic surface of the robot.

Don't Be Afraid of Applying Elbow Grease

Sometimes when I talk with founders, I always encourage them to apply elbow grease. In this case, it meant that I sat down on our assembly line and built an early production run of our robot by ourself.

I went through every step of our production process. From assembling gearboxes to screwing together plastic casings.

The great benefit of sitting down and building your own products is that you immediately see where the costs were increased and how you can make the process easier and better on the operators. And those changes were incorporated in every following production stage.

Practice Makes Perfect

Going through the various stages of Manufacturing
Our initial production run was the most gruesome one. I remember staying up with everyone until 2AM trying to get the first 10 units off the production floor. And even those products were terrible. There were scuffed up, not sealed properly, and there was even one that didn’t even work. But that’s why there are the various production stages.

After going through the EVT/DVT/PVT/MP stages, we were able to start producing units at a consistent speed with consistent quality. Admittedly, there were cases where we had to run certain stages multiple times because of missing parts, operator error.

But to quote one of my favorite movies, Inglourious Basterds:

“You know how to get to Carnegie Hall, don’t ya? Practice!”