DiscoverThe Hardtech Podcast
The Hardtech Podcast

The Hardtech Podcast

Author: The Hardtech Podcast

Subscribed: 7Played: 19
Share

Description

The Hardtech Podcast pulls back the curtain on the bold ideas, design challenges, and engineering breakthroughs behind today’s most innovative hardware products — and the people who make them happen. From first sketch to final build, we dive deep into the minds shaping the future of physical tech.
69 Episodes
Reverse
Enzo Jia was in Indianapolis for FDIC and offered to come record in person, which is how this conversation happened. He is President and Co-Founder of Longan Vision, which makes an AR and thermal heads-up display that mounts to a firefighter's helmet, and he joins DeAndre Harakas and Grant Chapman to trace what four co-founders out of the same McMaster University lab learned building something that has to work inside a burning building. The problem is easy to picture. A firefighter searching a dark, smoke-filled room is holding a thermal imaging camera in one hand, so one hand is gone. Longan Vision moves that view in front of their eyes and adds recording, streaming, and a way for a chief to push a message that sits in the firefighter's field of view rather than vanishing into radio chatter. The version history is the most interesting part. Generation one had every feature they could think of. Generations two through six were almost entirely subtraction, driven by fire departments telling them the weight mattered more than the feature list, and by the fact that every added feature generates heat inside a device whose job is keeping heat out. Off-the-shelf display optics did not survive the environment, so they went back to first principles and built their own. The controls tell the same story, since firefighters operate the device through heavy gloves, which is why it uses a rotary knob, why the knob got clickier over successive versions, and why changing modes eventually took two clicks instead of one. Enzo also describes a prototyping path worth copying, moving from high-temperature 3D prints for ergonomics to machined blocks for material testing, and only cutting tooling once users had weighed in. Longan Vision is about two years into the market now with 10 to 15 fire departments globally, including testing with the Tokyo Fire Department, a World Expo showcase, and work with a NATO partner in Europe, and the platform is moving into mining and oil and gas. He closes on advice for founders, and on why he is glad he did not know what was ahead when they started.Special Guest: Enzo Jia.
Software got its Google Docs moment years ago, the one where files stopped circulating as email attachments and everyone simply worked in the same place at the same time. Mechanical CAD followed. Electronic design took longer, and this conversation is about what changed when it arrived. Justin Sears is Head of Product Marketing for SaaS at Altium, and he came to electronics from a decade in big data having never designed a board or written code, which gives him a useful outsider's read on how hardware teams actually work. He joins DeAndre Harakas and Grant Chapman in person at Glassboard in Indianapolis, after a dinner in Silicon Valley hosted by Duro seated Grant at the end of the table and Justin at the head of it. Grant describes the change from the inside. Glassboard had hand-built a git repository to version Altium files, which mostly worked, and three months after moving to the cloud platform the design review meetings got shorter, the scheduled client calls disappeared, and the email traffic dropped, because the conversation moved into the tool where an engineer could answer in context instead of three days later. The traceability side matters just as much, giving a medical device client a design history file they can rewind to day one, and giving Glassboard a way to handle the roughly 20 percent of inbound work that amounts to fixing a board after a component went end of life. The two of them also work through where AI genuinely fits. Requirements are the right starting point because they have always been written down, while layout is a dark art passed between engineers and never documented, which is exactly the training data a model does not have. Grant argues for AI as a filter and a garbage collector that surfaces what deserves an engineer's attention, rather than something making changes on its own. The conversation closes on supply chain, notification burnout, why Glassboard calls itself the gateway drug dealer for Altium, and DeAndre's point that client visibility is really a retention strategy, the same reason the Uber map made waiting bearable.Special Guest: Justin Sears.
Serhii Zatsarynin spent years at Siemens Healthineers running in-vitro diagnostics in Ukraine, in a world where every human touching the process counts as a potential error, and then came home to find the most important test in his life, the one he and his wife were running while trying for a baby, was a strip in a cup read by eye. That gap is where Ovul started. In this in-person conversation with DeAndre Harakas and Grant Chapman, Serhii opens with a line that runs against the usual framing on this show, which is that hardware is not hard anymore. His case is that prototyping speed has changed what's possible, and he backs it with Ovul's own path, a working prototype in 30 days, pilot units with users about two months in, and roughly one iteration every two weeks after that. The product is a reusable digital saliva microscope that reads saliva ferning, a crystallization pattern tied to estrogen that was discovered in 1945 and has been sold for years as a manual device requiring a week or two of training to interpret. Serhii's move was to take the human out of that step with image recognition. The conversation covers the parts of hardware that surprise first-time founders, including early units that broke every time a user opened them, why a device designed to have no interface ended up needing light indicators, the painful stretch between a few working prototypes and fifty consistent ones, the discovery that you cannot injection mold the shape you 3D printed, and why Ovul chose Wi-Fi over Bluetooth despite the cost, because a device left at home has to sync without the user thinking about it. Serhii also talks about raising capital with product already in market and finding investors unsettled rather than reassured by it. The second half is about where users took the company, from ovulation detection into perimenopause and menopause transition, with around 360 serial production devices now in the field surfacing patterns the team didn't go looking for, and why Ovul deliberately collects no name, email, or phone number. He closes with a first distribution partner in Canada and the advice Grant calls the lesson of the episode, which is to go sell the thing even if it's ugly, because free feedback and paid feedback are not the same feedback.Special Guest: Serhii Zatsarynin.
This one is the peanut butter and jelly of connected hardware, a software consultancy that builds for hardware startups sitting down with a hardware development firm. Peter Tuszynski, CTO at intent, joins DeAndre Harakas and Grant Chapman to talk about what AI has actually changed about bringing an IoT product to life, and what it has not. Peter's answer is more grounded than most. He has seen individual engineers go from 1.5x to 7x on output, but he calls that unrealized capital gains, because shipping more code just means shipping more unverified code until the whole process gets the benefit of that speed. Grant's framing matches, that the real return on AI today is de-risking rather than velocity, since teams using it well are looking under rocks they would never have turned over, and every time you speed up one stage the bottleneck reappears further down the line. At intent the bottleneck turned out to be manual QA, because their projects involve a phone, a peripheral, and a cloud backend, and the messiest failures live between the phone and the peripheral where Bluetooth is low bandwidth and the device landscape is a mess. So they built a QA agent that is agnostic of the underlying model, wired up to drive iOS or Android, a tethered peripheral over CLI, and a test case repository, and found that newer models can read a human readable test case and work through it the way a QA engineer would. The conversation also covers why embedded was supposed to be out of reach for AI and what actually changed, which came down to giving a model a board to flash and a way to verify what it flashed, and from there to giving models control of lab equipment and closing the loop between the physical world and what a model can touch. Peter traces the progression from prompt engineering to context engineering to harness engineering to loop engineering, and the two of them get into designing hardware to be testable in the first place, the carpenter and the CNC machine as the analogy for engineers worried about their craft, why prompting has started to look more like talking than writing, and why play, not work, is how anyone actually figures out what these tools are good for. Filip Pietraszkiewicz closes with advice for founders on getting something clickable in front of users as early as possible.Special Guests: Filip Pietraszkiewicz and Peter Tuszynski.
Epitel just closed a $26 million Series B to expand remote wireless EEG monitoring, and Mark Lehmkuhle, the company's CTO and founder, and J Carroll, its Director of Software Engineering and Architecture, join DeAndre Harakas and Grant Chapman to talk through how the technology got out of the lab and into the world. Mark describes Epitel as a machine learning company for detecting neurological disorders that happens to build its own wireless wearable EEG sensor, a small chicklet-sized device with two gold electrodes that sticks to the scalp with a single piece of adhesive and conductive hydrogel. The hardware, in his framing, is the means to an end, because the product is really about converting a huge stream of data into clinically actionable information. The origin traces back to a decades-old NIH anti-seizure screening program at the University of Utah, where capturing rare seizure events in rodent models meant recording EEG around the clock for nearly the lifetime of the animal. Nothing existed to do that, so they built it, using a concept borrowed from the wireless tire pressure monitor in a car, with thumbnail-sized transmitters that could record continuously for six months. The conversation gets into the electrode research behind the current product, including recreating a study from the early 1970s to determine the minimum number of electrodes needed to still see the electrographic signature of a seizure, which came down to four locations, and a second study on how close together those electrodes can sit. Mark also explains why any of this matters commercially, since self-reporting remains the primary outcome measure in many trials and is roughly fifty-fifty at best. The rest is about the part Mark says is genuinely hard, which is human factors rather than the tech stack. Their users run from neonates to people in their 90s, seizure forecasting raises the question of what a percentage chance means to any individual, and a sensor nobody wears produces no data at all. J talks about garbage in and garbage out, why knowing a recording is bad beats not knowing, and why usability is a first class citizen on his engineering team. Grant and DeAndre add the lily pad analogy for small companies that cannot cross the pond in one jump, why Epitel's first FDA clearance was a milestone rather than a sellable product, why connected devices are never finished as long as they are supported, and why a little ignorance may be the only thing that gets a first-time founder to launch at all.Special Guests: Justin ("J") Carroll and Mark Lehmkuhle.
loading
Comments