Sematext: virtual enterprise IT product team with Park Slope founder

In these days of consumer web startup attention, it’s often the companies you’ve never heard about that are among the most profitable. So though the team behind Sematext has built machine learning into tool that lets enterprises more easily review the performance of its IT infrastructure, you won’t see founder Otis Gospodnetić on TV anytime soon.

Working out of the Brooklyn Creative League with one of his 12 team members, who are otherwise virtual, he’s doing what so many of those quiet enterprise IT firms do: finding new customers for his products by way of consulting and using revenue to slowly bootstrap its grwoth.

He started the company on paper in 2007 and started working on it exclusively full time in 2010. It’s been profitable from the beginning and never sought investors. It currently has a decentralized team of about a dozen people, all developers save one.

Here are a few of the products that Sematext has developed:

• SPM. Performance Monitoring and Alerting. This is where machine learning is important. It collects data from all kinds of sources, but its ability to learn is what makes it important. In many IT infrastructures, teams have to define anomalies manually — set rules or alerts for when something isn’t working right for a system. Instead SPM learns what a company’s anomalies are over time. So, for example, 60 percent server useage could be a giant spike early in a company’s history and normal later. SPM learns to tell the difference and alerts teams to moments that are truly unusual.
• Site Search Analytics. This is a way of looking at searches that are happening on your site more robustly than, say, Google Analytics can offer. Are customers doing several versions of a search with very similar language, because they aren’t finding what they are looking for? “Every day your search engine needs to search more,” Gospodnetić said. This is a utility that will let you know if your site is keeping up with the increased demand.
• Logsene. Servers and computers are constantly keeping logs of their activity. If a server farm goes down, an IT team might find what went wrong by checking the logs on each server, one by one, until they find the origin of the problem. Logsene copies all those logs into the cloud and puts them into a dashboard where teams can monitor performance more easily.

Gospodnetić pointed to two much larger companies that compete with Sematext‘s products: Splunk and New Relic. The former is more in the Logsene space and the latter is more in the SPM space. Gospodnetić argues that his products win easily on cost but they also win in user experience. By collecting all this information in one UI with a consistent methodology and underlying logic, companies get a more consistent feel for all their data across the different pools.

After all, the reasons for lagging search performance might be found in your server logs and performance bottlenecks could have something to do with what people are searching for, and so on. The relationships between these backend needs are what matter.

As a profitable company with a well developed set of software products, Gospodnetić said that Sematext would be a good candidate for outside investment, but so far he isn’t interested.

With VC money, the Park Slope resident said “You can go much faster, get market share, but then you have to do what VC’s tell you.”

He said the first thing that investors would tell him to do is completely cut off the consulting side of their business, where they help companies make their search systems work better.  The idea would be, with funding, hire a sales team and push to expand the footprint of their core products. Go big, get acquired or fail.

But the consulting is where he builds relationships and develops leads, he said. For now, that’s the right fit.

“The future will be in product,” Gospodnetić said, but he intends to grow it steadily, so that the company can sustain itself.  Gospodnetić wants to build a good business, not an acquisition candidate, so he’s content to take it slowly for now.