A new AI orchestration startup from the founders of Lithuanian unicorn Nord Security is setting out to help enterprises move AI projects into production, with an initial focus on bringing greater visibility, security and safety to large language models (LLMs). and adaptability.

The startup, called Nexos.ai, is the brainchild of Tomas Okmanas (pictured above) and Eimantas Sabaliauskas, who have built a brand that is one of the most recognizable not just in Lithuania, but across Europe. Nord Security is best known for its flagship VPN product, NordVPN, which bootstrapped the company for its first 10 years, eventually landing a massive $100 million investment in 2022 at a $1.6 billion valuation.

Today, their new company emerges from stealth with $8 million in funding from a host of high-profile backers, including lead investor Index Ventures, which has now made its first investment in Lithuania.

"We've known Tomas and the work he's done for many years, so when we heard he was building a new company in artificial intelligence and was ultimately willing to take venture capital at this (early) stage, Index Ventures partner Hannah Seal told TechCrunch.

Other notable investors include Creandum and Dig Ventures, as well as high-profile angel investors including the CEOs of Datadog, Klarna, Supercell and Wix.

Utilizing Catalysts

Currently, teams that want to put AI into production must connect countless tools, which can involve recruiting and building a team with the necessary skills. This is where Nexos.ai wants to step in.

“I see a big gap between AI pilots and going into production,” Okmanas told TechCrunch in an interview. “When you test AI in the lab, it might work and be useful, but when you want to put it into production, especially in the enterprise, how do you ensure high availability? How do you ensure security? How do you manage it? cost?"

Nord Security has been around for more than a decade, but five years ago it was folded into an umbrella company called Tesonet, an incubator with more than two dozen businesses. One of them is web hosting company Hostinger, which recently added AI-powered smart features to its website building tools. Okmanas, a board member and shareholder at Hostinger, said some of the problems they encountered became the catalyst for Nexos.ai's eventual development.

"We wanted to use artificial intelligence in our website builder, so we enabled OpenAI, started testing it, and put it into production," Okmanas said. "In August, we got a bill for $150,000. For what? Why is it so expensive? There's no visibility."

When OpenAI experienced several outages, Okmanas became convinced that something had to be done to make it easier to deploy, manage and optimize the "increasingly complex ecosystem of AI models" that organizations might need.

Through a simple API (Application Programming Interface), customers can access more than 200 AI models, ranging from well-known existing models like OpenAI and Anthropic to smaller niche LLMs. The idea is that if OpenAI goes down, the company can temporarily (automatically) switch to a different provider without a break in pace. Or, if for some reason the cost of obtaining a specific LL.M. surges, the company can switch to another LL.M. to lower costs.

Nexos.ai has also introduced "intelligent caching" - if a particular question is repeated by multiple users, the system can turn to its own database, rather than continuing to hire an LL.M., which can become expensive.

In terms of security and compliance, Nexos.ai also prevents individuals from sending private data to LLM providers, or their access can be terminated immediately if an employee leaves the company.

The elephant in the room, however, is unavoidable: One of the reasons businesses are hesitant to embrace AI is the thorny issue of data security—a healthcare company, bank, or insurance company cannot simply trust an LL.M. provider and all its sensitive information. It's worth noting that Hostinger itself suffered a data breach in 2019, and NordVPN has suffered hacking attacks in the past - attacks that all companies face today.

This raises questions about how Nexos.ai handles this data, given that it hosts all of its data on its own infrastructure. Okermanas said the company may offer self-hosted services in the future, and it already supports integration with the company's own in-house LL.M.

It also has guardrails in place to detect when data, such as personally identifiable information (PII), is sent to it - in which case it can reroute the data back to the original company's own LLM or database. But if the query is generic, such as a customer asking an AI agent for details about their location and opening hours, the query will be handled on the Nexos.ai side.

from idea to start

It took Nexos.ai about six weeks to go from idea to official launch, and while the speed at which it secured funding largely depended on the pedigree of the founders, a lot of it was just timing.

"I feel like we've finally moved beyond the hype of AI and now real-world applications are on the horizon," Hill added. "All the large enterprises are realizing that this really makes sense and they need to adopt AI at scale. It's time for the infrastructure to catch up to the model."

However, the speed of execution is largely due to Tesonet's broader organizational set-up, which has approximately 4,000 employees across its portfolio. This allowed Okmanas to quickly assemble a team of about 30 people he knew and trusted to work on Nexos.ai full-time.

“We have these teams that really come together — they’ve been working together for years and there’s no need to tell them what’s what,” Okermanus said. “We will also recruit from outside, but that will take more time.”

Nexos.ai's platform is set to launch at the end of March, but Okmanas said it's already working with a group of "test customers and design partners."