Going granular for data security in an A.I. world
Klarytee wants to give businesses peace of mind with smarter encryption
PreSeed Now brings you an in-depth profile of a different early-stage B2B or deep tech startup every Tuesday and Thursday. Subscribe to get it straight to your inbox.
It seems like every business wants to get into using A.I. these days.
But with Amazon’s business-focused chatbot having reportedly leaked confidential data, and ChatGPT having had a bug that got it to spit out un-redacted training data, it’s clear that security practices need to catch up.
Today’s startup might have a solution: granular, at-source encryption of data. And it’s not just A.I. use cases they can help with.
Scroll down to read all about Klarytee, which is today announcing it has raised a £700,000 pre-seed round.
Klarytee wants to get granular with data security in an A.I. world
Any smart business will take steps to protect valuable or sensitive data, even when it’s not a legal requirement.
But that data can leak, either by accident or by an employee’s deliberate action. It’s this problem that London-based Klarytee wants to solve. It offers granular control over how different data can be handled, with policies that can be automatically applied to specific data types.
“We build security into the data itself,” explains founder Nithin Thomas. “We can free organisations to be able to leverage things like A.I. and SaaS, and cloud and all of these things without having to worry about security, because the security follows the data wherever it goes.”
As an example, Thomas explains how a bank could use the technology to add granular encryption into customer names, pricing information, and other sensitive data it would want to protect.
“Once we build that into the document, wherever that data then goes, even if you copy it out of the document and you put it into a CRM, database, or you copy the document and so on, the sensitive bits of that data will stay encrypted.”
This means access control rules will apply so that, for example, sensitive data won’t be included in a printout. Thomas says this also means that companies are able to share data with large language models without worrying about sensitive data accidentally being shared.
“There are also other use cases,” he adds. “Like for instance, our end goal is one day if you're a retail bank, if your customer forgets their login credentials, they should be able to just message you on social media and you are able to just message them back with the password in encrypted form.
“And because the data itself is protected, you don't care whether the online platform is secure or not.”
How it works
In practice, Klarytee works by integrating with the software that companies already use, and it already offers a Microsoft Word integration. Once they’re set up with a subscription, organisations can set rules for how different types of data in documents should be handled.
Say a business decides that customer names can only be accessible by company staff members. This sensitive data can then be automatically detected in a document by the software, or selected manually by the user.
Those customer names then get immediately encrypted in a way that means that when the document is opened, only someone who authenticates themself as a company employee can reveal them.
The decrypted data is permanently erased from memory once the app or device is closed, meaning authentication is required every time the document is accessed.
Thomas says Klarytee’s tech can automatically detect things like passport numbers, credit card information, addresses, and the like. He says it’s based around a local large language model trained on relevant data, which has the potential to become even smarter.
“As we gather more and more insights into the data, what we're building is more sophisticated models where we can also look at the context behind the document. So we can say ‘this document looks like a contract’. Maybe it's got some pricing information in it. ‘Are you sure you want to leave that information unprotected?’”
Thomas says he wants to give “powerful prompts to the user based on the context of their data, so that they can then make decisions on how to handle that data.”
The story so far
Thomas has a PhD that was focused on data encryption. After completing it in 2010, he went straight into creating his first startup, SQR Systems, which was based around his research.
“I was researching new ways of encrypting data for very poorly connected networks… I worked with the defence and national security community around deploying that for some very sensitive government applications,” he says.
Through this work, he saw how security practices were often a frustrating barrier to productivity, and thus the idea for a business-focused startup was born. Thomas is the sole founder of Klarytee, which is now a team of six.
Earlier this year, they took part in Accenture’s FinTech Innovation Lab programme, which Thomas says helped get Klarytee moving forward by allowing them to validate the tech with international tier-one banks.
“From day one, we were working very closely with our customer base, to make sure that we’re building something that addresses real problems. As a second-time founder, I was very sensitive to that, to make sure that we build something that addresses a real problem, and not just a technology looking for a solution.”
With an eye on ensuring that the encryption is more of a benefit than a hassle, Klarytee is working to improve its A.I. data detection technology.
“Our main focus right now is around automating the way we do the encryption,” says Thomas.
“We're building more and more A.I. technology into our platform, so that instead of relying on the user to be able to identify sensitive data, we can automate all of that and make it really seamless for the user so that the encryption gets out of the way completely.
“We still allow the user to have control over their data, but actually, we're using A.I. as an enabler to simplify a lot of those workflows.”
Klarytee has been bootstrapped until now. It has just announced that it has raised a £700,000 pre-seed funding round.
The round was led by Concept Ventures, alongside angels like former Twitter CISO, Micheal Coates; former group CEO of Bupa, Evelyn Bourke; the founders of cyber security company Digital Shadows; former managing director of Accenture, Brad Cable; and Desigan Chinniah, board member of the Tor Project.
“I was very selective about who I wanted in this round, because I wanted strategic money early on and people that can give me access to these big platforms where I can find integration partners,” says Thomas.
“We want to build integrations deep into the existing enterprise tech stack, instead of giving our customers something that's completely standalone. We want to be completely embedded into the tools that they already use. So investors that have access to those companies are really important for me.”
Go deeper on Klarytee
Read more about their competition, vision, and the challenges they face: