Coolan lets companies pool and analyze hardware data

A common dilemma facing many companies that have a ton of gear in their data centers is having to figure out which hardware appliance is causing bottlenecks that may cause downtime and customer outrage. Coolan, a startup formed by former Facebook and Google engineers, aims to solve this problem and is exiting stealth with a new product that gathers together infrastructure data from multiple companies, which it then analyzes to unearth how all their gear is performing.

That strength-by-numbers approach separates Coolan from other IT monitoring services out there that companies plug into their data centers to discover how efficient (or not) their infrastructure really is so they can spot problems before they turn to something bigger.

Although these IT monitoring services essentially study the infrastructure of one company, Coolan’s software platform allows other entities to share their infrastructure data with each other in the hopes that, with more data available, organizations can put an end to unnecessary server failures and the like.

“[Organizations] are all curious about solving this problem, but they have a limited data set,” said Coolan co-founder and CEO Amir Michael. “By bringing the industry together you get a larger data set.”

Screenshot of failure rate

Screenshot of failure rate

Michael was a a hardware engineer at [company]Google[/company] and then a Facebook engineer and manager of the company’s hardware design. At [company]Facebook[/company], Michael’s contributions led to the creation of the Open Compute Project, where he is is still an active participant as vice-chair of the project’s Incubation Committee, responsible for reviewing new specifications.

The open compute project did a good job of getting people to talk about servers and design concepts from a hardware level, Michael said. However, when it comes to operations and getting the most out of hardware, there’s not a lot of information available to the general public on the actual performance metrics of individual pieces of hardware.

“We all want more transparency around our hardware,” Michael said.

The idea is for companies big and small, with 100 servers or 1,000 servers, to all benefit from the insights gleaned from the same big data set. Companies will have to install software (three lines of code, apparently) onto their fleet of servers, which will allow for infrastructure data to flow over to Coolan’s own servers, stored in Amazon S3.

Screenshot of notification report

Screenshot of notification report

Michael seemed aware of the irony that his startup that specializes in hardware-performance metrics operates in the cloud, but he said that “we will eat our own dog food and be running our own servers” once it reaches a certain size.

Coolan will not be syphoning the type of software-related data that New Relic or AppDynamics need for their analytics purposes, but rather hardware data, like the name of a device manufacturer, the temperature of the hardware when running, the model number of an appliance, when the device started generating errors, and so on.

From all this data, Coolan’s team can run machine-learning algorithms to learn how the hardware stacks up and which devices have a higher chance of failure. If a bunch of companies that contribute to Coolan all find that a fan in a particular manufacturer’s device cracks out at the two-year mark, then users who recently purchased that device will now have some warning that their devices might not function properly down the road.

Coolan CEO Amir Michael

Coolan CEO Amir Michael

Coolan’s not ready to disclose who its pilot customers are, but Michael did say that a number of the company’s clients are organizations that started out in the cloud and are now moving off of it to build their own data centers.

The startup could one day have a tool that also monitors a company’s cloud infrastructure, but Michael said that’s “not the primary focus right now.” Coolan is also still figuring out its pricing model, but its main goal as of now is to simply get more companies on board to “get more data.”

“I think part of it is in my DNA,” said Michael in reference to how his days at Facebook could have made him more open to the idea of sharing and collaborative projects. Facebook recently launched a collaborative threat-detection framework that seems similar to Coolan except instead of hardware data, companies are dumping into a central hub security data.

The six-person team at Coolan is not disclosing how much funding it has raised so far, but it closed a seed round in February led by Social + Capital, North Bridge Venture Partners and Keshif Ventures.

Come see August, Electric Imp and more at our SXSW Hardware House

No matter what you think about South by Southwest Interactive, tens of thousands of people still head to Austin, Texas in the Spring to partake of tacos, barbecue and some decidedly weird tech culture when the second and third weeks in March collide. So this year Gigaom and Stage Two have teamed up to throw a celebratory happy hour to honor those who love and build gadgets and connected devices.


On Friday March 13th we’ll have an evening event — The SXSW Hardware House — featuring interviews with Jason Johnson, CEO of August Locks; Hugo Fiennes, CEO of Electric Imp; Sam deBrouwer, Co-founder of Scanadu; and Nick Yulman community manager for hardware and design at Kickstarter, all prepared to share their insights and tips about getting your products off the ground and into consumers hands. They’ll have stories to share about manufacturing, crowdfunding, government regulations and finding the right retail partners.

Between each of these interviews we’ll also have demonstrations from some of the most exciting products out on the market, and some that haven’t even launched yet from companies like Leeo, OMsignal, MetaWear and more. We’re still looking for some demos, so if you are interested fill out this form by the end of this week and we’ll evaluate the applications and let you know if we have room for your demo.

The event kicks off at 6 pm at the WeWork space at Sixth and Congress Ave. at the heart of the SXSW action. We’ll have some space set aside for tables and demos on the first floor for smaller startups and then an elevator ride up we’ll have the main event with tacos, beer, the presentations themselves and even live music for the full South by experience. If you’re into hardware and the internet of things, I hope to see you there.

Want to attract the average consumer? Skip the hardware preview

For the last two years I’ve been reviewing connected devices as part of my role as Gigaom’s internet of things reporter. I’ve spent hours fiddling with radio networks, sliced my fingers open attaching wireless hubs to my garage door opener, giggled with my daughter while weighing out sugar on a connected scale, and generally had a good time. I’ve even had a few adventures in home wiring and and the occasional shock.

But of the dozens of products I’ve tested, only a few have actually been ready for the market when they’ve reached my hands. In a few cases this is to be expected: some were Kickstarter-backed projects that were shipping to the developers and reviewers at the same time and some were hitting my doorstep a month in advance of their availability in stores.

But others were shipping to the general backers or worse, in stores as I was testing my own versions of the products. And they just weren’t ready.

After two years of this, I think it’s time the market and its participants stop shipping products that aren’t up to snuff. Because as more people start picking up these devices and trying to see what the smart home is all about, a bad experience with an underperforming or buggy product is going to turn them off from the whole concept.

Sometimes, it’s worth the wait

Take for example, the Wink home hub from Quirky. This product launched in July with major backing from Wink and Home Depot. My guess is the companies involved felt the pressure to get into the market quickly since there were already several players ranging from small (SmartThings, Revolv) to big (Staples Connect and Lowes Iris) that had been selling for months and even years in the case of Lowes.

But when Wink launched, it was buggy, lacking support for basic devices that it shared radios and aisle space with, and consumers who picked it up were pretty frustrated. Luckily, it was cheap, and Wink had the cash to keep it in the market and manage its retailers’ frustration with returns — a luxury smaller startups don’t generally have. But still, any customer who decided to take a chance on Wink to see what the fuss was about wasn’t likely to walk about thrilled with the smart home experience.

Wink wasn’t ready for market. But what about LIFX, the $99 connected light bulbs that change color and connect via Wi-Fi? The bulbs were pricey, but since Philips Hue had primed the market, others were coming in with newer variations, and LIFX was winning accolades. But in July, after a few of my friends had picked up the bulbs, an English hacking firm reported that the bulbs exposed users to a significant security flaw by letting people hack into to users’ Wi-Fi networks. Not only that, LIFX stored users Wi-Fi passwords all in one database. Both problems were quickly solved, but the exploit and the database setup exposed some basic flaws in how LIFX had handled security and notifications. Simply put, it wasn’t ready.

Take it on faith

My final beef with products that aren’t ready concerns those companies that are pitching some type of algorithm as part of their value, but the algorithm isn’t quite trained on enough data yet. Much like the vaporware of the 1990s, I think of these products as faithware, as in we just have to have faith that they’ll work well and do what we want them to do. Algorithms aren’t magic. They have to be trained and even the act of training them introduces biases that mean it may not work in the way we want it to.

[pullquote person=”” attribution=”” id=”913890″]I think of these products as faithware, as in we just have to have faith that they’ll work well and do what we want them to do.[/pullquote]

If something has a learning algorithm that assumes you love things one way (sleeping at night, for example) and you don’t live that way, it’s not going to make your life easier. So products like Canary home security system (which is on sale now and will ship in March) that offer an algorithm that will arrive eventually worry me. If learning is a big reason you are buying a product then make sure that’s a feature that exists straight out of the box when you buy it rather than down the road. Because ideally you’ll want to train it within your return window for that item.

The home is a shared environment

This may sound harsh. I know that all of these companies are after a first mover advantage and everybody is out there right now popping a Wi-Fi or Bluetooth module on an everyday object and building an app for it. But hardware is not software. There’s only so much iteration a consumer is willing to take when it comes to a functional object that she plans to install in her home. Especially an object that will be used by everyone in that home.

Nest and Philips Hue both were consumer ready products.

Nest and Philips Hue both were consumer ready products.

It’s fine when your app crashes; at worst, you reboot and move on. But when you spend $250 on a connected door lock that your husband was a bit dubious about in the first place, and he comes home with his arms full of groceries and gets stuck on the doorstep because the lock didn’t open, you can’t tell him to reboot. That lock is going to get returned. And my bet is your mission to connect your home just got set back by a few years.

As I’ve been playing with devices I’ve expected the quality to get better over time. And while I think there has been a slight uptick in overall quality — devices coming out of the PCH incubator for example tend to arrive consumer-ready (although without as great support for Android devices) I’m still surprised at the overall uneven quality of the products that are shipping to consumers. Maybe some of the entrepreneurs who participated in how we built it campaign at our Structure Connect event last year can offer a lesson or two.


Please, guys. I love hardware and I love connected devices. But the mainstream consumer isn’t going to love anything that they have to spend hours troubleshooting. Or that they have to worry about from a security standpoint. We may love playing with our connected outlets and sensors, but trust me, everyone else wants to just set them and forget them.

You have to release products that let them do that. Until you can, keep them to yourselves.

Y Combinator doubles down on hardware startups with new resources

Citing a growing number of investments in hardware companies, Y Combinator announced today it will integrate two prototyping labs and expert hardware partners into its startup accelerator.

Y Combinator will build a small electronics shop in Mountain View, but the bulk of hardware work will happen at Autodesk’s Pier 9 space. Pier 9 is home to millions of dollars worth of equipment for metalworking, woodworking, 3D printing and even biochemistry.

Just one room in the massive Pier 9 prototypnig space.

Just one room in the massive Pier 9 prototyping space.

The accelerator’s startups will work out of Pier 9 with the help of Bolt, a venture capital firm that specializes in product design and manufacturing. Other prototyping studios and past Y Combinator startups are offering up further expertise, service discounts and equipment.

Hardware startups must take a more arduous path than new software companies, as it is a long and expensive process to go from idea to prototype to product. Hardware accelerators like Highway1 and Lemnos Labs have cropped up in the Bay Area in recent years to provide startups with basic tools and help them locate resources and factories in China.

As the internet of things begins to meld hardware and software and Silicon Valley grows more interested in more challenging “moonshots,” formerly software-centric accelerators like Y Combinator are taking notice. President Sam Altman wrote in a blog post that Y Combinator will be putting out requests for more hardware startups.

“We don’t shy away from expensive hardware,” he wrote, citing Y Combinator’s investments in startups like UPower, which focuses on nuclear fission. “We’re happy to see all sorts of hardware companies, but we especially like the ones that are fundamentally new ideas that Kickstarter might not support.”

TeraDeep wants to bring deep learning to your dumb devices

Open the closet of any gadget geek or computer nerd, and you’re likely to find a lot of skeletons. Stacked deep in a cardboard box or Tupperware tub, there they are: The remains of webcams, routers, phones and other devices deemed too obsolete to keep using and left to rot, metaphorically speaking, until they eventually find their way to a Best Buy recycling bin.

However, an under-the-radar startup called TeraDeep has developed a way to revive at least a few of those old devices by giving them the power of deep learning. The company has built a module that it calls the CAMCUE, which runs on an ARM-based processor and is designed to plug into other gear and run deep neural network algorithms on the inputs they send through. It could turn an old webcam into something with the smart features of a Dropcam, if not smarter.

“You can basically turn our little device into anything you want,” said TeraDeep co-founder and CTO Eugenio Culurciello during a recent interview. That potential is why the company won a Structure Data award as one of most-promising startups to launch in 2014, and will be presenting at our Structure Data conference in March.

Didier Lacroix (left) and Eugenio Culurciello (right)

Didier Lacroix (left) and Eugenio Culurciello (right)

But before TeraDeep can start transforming the world’s dumb gear into smart gear, the company needs to grow — a lot. It’s headquartered in San Mateo, California, and is the brainchild of Culurciello, who moonlights as an associate professor of engineering at Purdue University in Indiana. It has 10 employees, only three of which are full-time. It has a prototype of the CAMCUE, but isn’t ready to start mass-producing the modules and getting them into developers’ hands.

I recently saw a prototype of it at a deep learning conference in San Francisco, and was impressed by its how well it worked, albeit in a simple use case. Culurciello hooked the CAMCUE up to a webcam and to a laptop, and as he panned the camera, the display on the computer screen would alert the presence of a human when I was in the shot.

“As long as you look human-like, it’s going to detect you,” he said.

The prototype system can be set to detect a number of objects, including iPhones, which it was able to do when the phone was held vertically.

teradeep setup

The webcam setup on a conference table.

TeraDeep also has developed a web application, software libraries and a cloud platform that Culurciello said should make it fairly easy for power users and application developers, initially, and then perhaps everyday consumers to train TeraDeep-powered devices to do what they want them to do. It could be “as easy as uploading a bunch of images,” he said.

“You don’t need to be a programmer to make these things do magic,” TeraDeep CEO Didier Lacroix added.

But Culurciello and Lacroix have bigger plans for the company’s technology — which is the culmination of several years of work by Culurciello to develop specialized hardware for neural network algorithms — than just turning old webcams into smarter webcams. They’d like the company to become a platform player in the emerging artificial intelligence market, selling embedded hardware and software to fulfill the needs of hobbyists and large-scale device manufacturers alike.

A TeraDeep module, up close.

A TeraDeep module, up close.

It already has a few of the pieces in place. Aside from the CAMCUE module, which Lacroix said will soon shrink to about the surface area of a credit card, the company has also tuned its core technology (called nn-x, or neural network accelerator) to run on existing smartphone platforms. This means developers could build mobile apps that do computer vision at high speed and low power without relying on GPUs.

TeraDeep has also worked in system-on-a-chip design for partners that might want to embed more computing power into their devices. Think drones, cars and refrigerators, or smart-home gadgets a la the Amazon Echo and Jibo that rely heavily on voice recognition.

Lacroix said all the possibilities, and the interest it has received from folks who’ve seen and heard about the technology, are great, but noted that it might lead such a small company to suffer from a lack of focus or perhaps option paralysis.

“It’s overwhelming. We are a small company, and people get very excited,” he said. “… We cannot do everything. That’s a challenge for us.”

Not so handy? There’s a connected device for you, too

I love the idea of the internet of things, I really do. But as a renter I don’t want to buy a Nest or any other device that I would need to wire into the wall. Plus, my building is really, really old. I have a feeling it would be an arduous process.

That’s at odds with the internet of things’ big promise: simplicity. Connected devices are supposed to streamline our lives, but that doesn’t matter if we can’t figure out how to integrate them into our homes.

“Those are huge friction points,” said David Genet, co-founder of connected peephole maker Building 10. “If you can’t install it without calling tech support, you’re not going to buy it.”

Peeple attaches into existing peepholes.

Peeple attaches to existing peepholes.

I spent Tuesday morning at the third biannual Highway1 demo day and noticed something unusual: a group of connected devices that don’t require any rewiring at all. There was Lagoon, which straps onto homes’ main water lines to track usage. Peeple, a connected peephole that snaps a picture when someone knocks, fits into the existing hole in your door. Switchmate snaps over any light switch. Oh, and Fishbit, the connected water monitor, just drops into your fish tank.

No wires calls for a battery

So, why were we wiring devices into our homes in the first place? One reason is power. Those cords carry sweet, sweet electricity to the electronics, meaning they never have to be charged. All of the devices at the demo day need to be charged every 6 to 12 months.

If 6 to 12 months sounds like a long time, that’s because it is. Only recently did it become possible to build battery-powered products that can collect data and communicate it without frequent charging.

The Lagoon water monitor wraps around existing water meters with a strap.

The Lagoon water monitor wraps around existing water meters with a strap.

The Highway1 startups started with very little physical movement. There are no motors or moving parts to suck battery life. Then they added low-energy communication: Peeple uses Wi-Fi, Switchmate relies on Bluetooth Low Energy, Fishbit is compatible with both Bluetooth Low Energy and Wi-Fi, and Lagoon’s device comes with an unnamed radio technology. As my colleague Stacey Higginbotham has outlined extensively, this will be more and more common.

The future needs to be wired (or something totally different)

OK, so, you have your connected peephole. Now imagine a few years from now, when you also have connected light switches, water monitors, appliances and so on. That’s a lot of charging.

The Switchmate device sits on top of traditional light switches.

The Switchmate device sits on top of traditional light switches.

Switchmate CEO Robert Romano acknowledged that would be annoying, but added that most people are not yet connecting their entire homes. In these early days, people might adopt two or three devices for areas of their home that are especially important to them. That’s not such a big commitment.

When we do start connecting a larger portion of our homes, we might have to suck it up and do those often difficult installs. Or we might have some totally new ways of providing power. Romano suggested solar. Or, even better, we could skip ahead to wirelessly charging everything in our homes.

Until then, I’ll embrace the battery-powered connected devices trickling into the market. I know my landlord will thank me.