A pair of Airbnb alums is bringing intelligence and automation to data protection

When Julie Trias and Elizabeth Nammour were working together at Airbnb on the company’s data team, they had to deal with data spread across a variety of sources, and that growing sprawl led to challenges in keeping data safe. The founders’ own frustration with the existing crop of data protection options motivated them to launch a company and build the automated data protection tool they wanted.

On Tuesday, that startup, Teleskope, announced a $5 million seed investment.

“We tested a bunch of different tools to help us understand, protect, delete and redact data at Airbnb, but what we came to realize is that there wasn’t that tool that could help developers do this automatically,” Trias told TechCrunch.

That’s not to say there were no solutions, but the ones that existed like data classification tools generated a lot of false positives and had scaling issues. “There wasn’t a tool that could help you go from detection to actual remediation, which is redacting the data, isolating the data, or taking any sort of action on the data.” The solution Teleskope has built enables customers to connect to their various data sources, identify sensitive data across a variety of data stores in an automated way, and isolate or delete it when necessary.

They currently have a few different use cases: “We’re mainly now selling to data teams, not just a product developer, but data governance engineers, who want to clean up their entire data sets in their data warehouse, or they want to clean one data set before they use it for model training, or they have multiple data sets, and they need to delete data for a particular user for compliance purposes,” she said.

The solution relies on what Trias calls “a pipeline of models” with different ones coming into play, depending on the type of data. “So for example, we’ve trained a model that’s really good at classifying data in natural language like conversational types of files. We’ve trained a model that works really well with structured tabular types of formats. We’ve trained a model that can classify sensitive data in a code base file or a log file,” she said.

Trias says that in spite of having the background and pedigree to build a product like this, they weren’t well versed in the world of venture capital and how to pitch when they first launched the company — and female founding teams face a bigger challenge with investors in general. “I think the hardest part was that when we first started making VC calls, we had no idea how to go about it. We didn’t even know what a design partner was. We were pre-product, pre anything, and we didn’t know all the VC lingo. And so we were very unprepared when we first took our first meetings with VCs,” she said.

They refined their presentation over time, and were able to find investors who got them and their vision. The seed funding was led by Primary Venture Partners with participation from Lerer Hippeau and Plug and Play Ventures along with Essence VC, which led the company’s pre-seed round.