IoT API Startup Crowsnest to Offer Crash Analytics Service

Internet of Things API startup Crowsnest has pivoted to sharpen its focus on providing device analytics for IoT makers.

Originally focused on creating an open source platform to integrate IoT devices through a standardized set of APIs, the startup founders and Techstars alums have pivoted to create a crash analytics machine learning service for IoT device makers.

Crowsnest home page

“We interviewed over 100 Internet-connected product companies, and the signal we found amongst all the data we collected was that diagnostics was outside their core competency,” says Crowsnest co-founder Michael Kruk. “They need in-depth analytics to get to the root cause of device crashes faster. And it is something they would be willing to pay money for and was outside their core business.”

Kruk says these interviews helped the Crowsnest team identify the value opportunity it could provide in the emerging IoT landscape. “Another big key learning was that a lot of companies were trying to do this analytics themselves, but for the data ingestion of the massive amounts of logs this created, they required a special hire, whether that be a machine learning engineer or a big data scientist.”

Kruk believes Crowsnest was perfectly positioned to pivot and build the crash analytics service on the back of Crowsnest’s original approach, even though this required building a new tech. “The technology is the strength of our team — to build tech quickly and effectively. We were able to maintain a lot of the same relationships: companies that were interested in working with us in the past. A lot of those companies came through the pivot and helped us with the new idea.”

Analytics dashboard

Kruk explains the core of the service for any device maker that has distributed its IoT products to end consumers:

Once you log in to your Crowsnest account, you have access to three tiles that can monitor thousands or millions of devices in customers' homes.

The first tile is the percentage of units that are having critical problems. The second tile shows you the three biggest bugs that would solve most of your problems, and the third tile compares your current crash rate with recent performance across the past two weeks.

In addition, Kruk says analytics divides issues based on severity as either critical or those that are more of a warning of poor performance, as well as providing information and debug data. A key focus has been on providing trends of the events that proceed crashes. “For example, we could notice that 87% of the time, before one device crashes, it involved someone checking for a firmware update. So this event trend data turns out to be a good way to search for the root cause,” says Kruk.

While Crowsnest’s tech has been built from the ground up to apply machine learning to the IoT market, APIs remain at the center of how Crowsnest works. A customer information API allows hardware makers to link devices with customer details such as email addresses so that they can automatically alert customers to any performance issues. The idea is that by helping device makers be proactive about crash and performance issues, they can provide a better customer experience and retain their IoT customer base. Kruk sees further API integrations as next in line to help clients link crash analytics directly to their CRM systems to further enable the use of analytics reporting as part of their proactive customer support culture for IoT devices.

An open packet specification API is also available so that potential Crowsnest customers can link their current data-logging procedures into Crowsnest’s system and UI. This allows those testing the service to see if they gain any additional value by being able to use the Crowsnest interface to analyze and track crashes without having to restructure their existing logging workflows to funnel through the Crowsnest’s product. They can continue to log their crash and performance data using their current processes while also making use of the API to duplicate the data into their Crowsnest accounts. Here, the company is using an API approach to enable customer acquisition by allowing its users to compare their current systems to what added value they may gain from using the Crowsnest dashboard product.

Kruk shares the road map for Crowsnest’s next features:

This is what we have today; this is phase one of the product. Next, we are really excited to be building out our alerts system. For example, if a particular type of crash happens, say, more than 10 times within a certain period of time, then you can set up alert emails or automatically roll back your hardware to a previous, more stable firmware update.

So at this point, we are doing correlation analysis so we have a way of identifying what is happening proceeding crashes, and then we will build more predictive analytics. That will let us start looking for patterns going forward, more of a machine data learning type approach.

Kruk sees Crowsnest benefiting early-stage companies that may have launched a thousand beta units and want to see how those are performing in their customers' hands. For bigger-value device manufacturers — companies with tens of thousands or even millions of products being shipped into very different environments — then the main goal is to help these customers retain their customer base as the IoT competition heats up. For now, the startup is focusing on consumer devices including wearables and home automation, with home automation hubs representing a particular opportunity. “We are monitoring close to 10,000 devices now,” says Kruk.

Developers and hardware makers can sign up for trial (beta) access on Crowsnest.

Mark Boyd is a ProgrammableWeb writer covering breaking news, API business strategies and models, open data, and smart cities.

Comments