The Lithuania AI Wave: Small Teams, Global Wins

The global pace of AI company development and implementation is accelerating and isn’t leaving Lithuania behind. Vilnius feels energetic and restless right now. A growing number of startups and established tech companies are leaning hard into AI to capture not only global market attention but customer market share.

I had the chance to interview a half dozen companies in Lithuania as part of their Going Global event, with company offerings spanning agentic AI, edtech, security and governance, healthcare, and enterprise offerings. Each company focuses on delivering real work rather than AI for AI’s sake. Each also sells outside Lithuania. Together they hint at a pattern that matters for a small market hungry for scale.

Agentic AI Moving Fast

Focusing on small- and medium-sized business (SMB), Sintra.ai offers task-oriented agents that are positioned as cute-looking avatars wearing different space-style suits and helmets. A user connects accounts, picks a goal, then watches a set of agents plan, create, and publish across channels.

For example, social media-focused agent “Soshie” provides a conversational task-oriented agent that connects with social media accounts to automatically craft and post media on behalf of SMB users. “We’re focused on the end to end workflow,” the team told me. They hear the same request from owners who juggle too much. “Give me one click. I just want to turn the social media on and forget about it.”

The company is growing fast. After a public launch in May, the team crossed one million dollars in annual revenue run rate (ARR) in under 60 days, and is currently on a track to do around $8M ARR. The company recently closed a $17M seed round to further accelerate growth, with a team of around 50 people.

The market for agentic AI solutions is highly competitive with new entrants emerging in the market on a daily basis, so the Vilnius-based company has their work cut out for them. The rapid growth in revenue and customer acquisition is a sign that there’s plenty of demand for everyone at the moment.

Providing College Admissions Help and Guidance to US and International Students

Vilnius-based firm Unive targets a gap that is becoming increasingly pressing for students applying to colleges or looking for higher education advice.The American School Counselor Association recommends a 250 to 1 student-to-counselor ratio, but the reality is that the ratio is much higher. The national average in the 2023–2024 year sat at 376 to 1. That math means teenagers make high-stakes choices with little timely advice. Counselors handle bloated caseloads and stray tasks that have little to do with admissions strategy. Many students report regret by graduation, and only 27 percent end up working in a field directly tied to their undergraduate major, according to the New York Fed’s long-running analysis of major-job match.

Unive aims its AI-powered solution as guidance that spans search, list building, essays, interviews, and portfolio strategy. “We specifically say guidance is a problem,” said company founder Jonas Kavaliauskas. “In the United States, you have 385 students on average per one guidance counselor. In Europe, a lot of countries don’t have this position almost at all.”

The startup’s solution runs seven agents that walk a student through selection and execution. The system pulls structured facts like tuition and rankings, mixes in qualitative signals, factors in student preferences, then constructs a reach-match-safety portfolio before moving to extracurricular framing and interview prep.

Unive’s wager is that a persistent, context-aware agent can move the decision quality up for families who cannot afford private counselors. The company plans a deeper high-school engagement that starts in ninth or tenth grade with psychometrics, early career exploration, and real-world trials like curated internships, then flows into applications.

“We have 200 customers out of the U.S., mostly juniors and seniors,” Kavaliauskasr said. Pricing is subscription based with monthly, quarterly, and annual plans while the team experiments with packaging. The team’s thesis follows a hard labor truth.

Continuing to Innovate AI in Radiology

Lithuanian-based Oxipit is focusing on an aspect of AI that has long had deep interest in demand, using AI to assist with analysis of medical imagery. Oxipit’s ChestLink suite holds a CE Class IIb mark in Europe for autonomous handling of normal chest X-rays. When the model is highly confident a study is normal, a radiologist can skip it and spend time elsewhere. That move addresses increasing radiologist throughput, not just accuracy.

When Oxipit tested 2,000 images with ten real-world radiologists, it received ten different reports for many cases. Agreement was only there when lungs looked clearly normal. That observation nudged the product toward autonomous normal-case clearing rather than opinionated abnormal reads.

“What you need to achieve is bigger than 99.9 percent sensitivity,” said company co-founder Vytautas Naujalis. The important thing is to focus on minimizing false negatives, since not spotting something of concern is more of an issue than incorrectly flagging something as concerning. Such normal X-rays can be up to 40% of all images, depending on the site. Separate detectors exist for major pathologies such as nodules, consolidation, and pneumothorax. The system even checks whether the input is a chest X-ray, not an incorrect image or a stray upload.

The key to the company’s offering is their rich database of real-world imagery data. “We now have more than two million chest X-rays,” Naujalis said, across Europe, South America, Asia, the United States, Australia, and parts of Africa.

Regulatory context helps explain why the use of AI in radiology still feels rare. Annalise Enterprise CXR, a leading chest X-ray suite, holds MDR Class IIb clearance as an assistive decision-support tool. It flags up to 124 findings. It does not let clinicians skip the read. In stroke care, Viz.ai won early FDA de novo clearance and now runs in more than 1,500 U.S. hospitals for triage and care coordination. Again, the design notifies specialists faster. It does not remove a clinician from the loop.

Oxipit’s team is tiny, with just 18 people, the founder said. The company reports wins across Scandinavia, including Copenhagen’s regional tender, plus traction in Norway and pending work in Estonia, Poland, and the UK. Brazil granted ANVISA clearance, and Australia has run deployments since 2019. The group is pursuing FDA clearance with a defined U.S. study pipeline and says a line of interested sites already waits for the green light.

Enterprise-Focused AI

Exacaster, another Lithuanian company pursuing rapid AI-powered expansion, predates the current agent craze by a decade. The company started with predicting customer churn in telecom models, then scaled into customer data management and enterprise AI without outside capital.

The founder’s early stories double as a thesis. He once coded a roulette system, saw “infinite returns,” won for a week, then re-read the code and found a bug. Later, a Wi-Fi drop erased a leveraged forex position. “I need to apply these algorithms in situations that would start creating value,” company co-founder Egidijus Pilypas decided. That pivot led straight to telco churn and a steady expansion into revenue-linked analytics.

Now AI is radically shifting the company’s way of doing business. “We will not do a single task ourselves without AI,” Pilypas told me, describing a company-wide shift six months in the making. He built the EXO DPM agent, a working copy of the Director of Product and Marketing wired into Exacaster’s business context. Proposal drafts, positioning, advisory “voice,” and sales support run through that agent layer. Under the hood sits a developer-first toolchain that borrows from Google’s Cloud Code and similar stacks to keep agents close to code rather than a brittle prompt zoo.

Exacaster ships like a hundred-person firm that treats agents as an operating system. Every new process starts with “how would the agent do it” rather than “who can we assign.” The company’s history with measurable telco outcomes gives the sales motion teeth.

How Lithuania is Crafting an AI Advantage

Lithuania’s edge does not flow from deep government subsidies or a fat venture funnel. It comes from founders who ship with small crews, pick narrow problems, and let software do the heavy lifting. Teams move fast, keep payroll light, and chase revenue in weeks, not years. AI fits that mindset since it replaces headcount creep with orchestration, data plumbing, and agents that finish work without hand-holding.

This orientation shows up inside the companies themselves. Leaders set a high bar for their own habits before preaching transformation to clients. As one founder at Exacaster put it, “we decided that we will not do a single task ourselves without AI.” Strip ceremony. Automate first. Then scale the pieces that measurably change the workday.

The rest follows. Sintra focuses on one-click outcomes for owners who have no time to wire tools together. Unive tackles guidance where counselors are outnumbered and families need a steady hand. Oxipit clears safe slices of radiology volume with confidence thresholds that pass a regulator’s test. Different buyers. Same posture. Small teams looking for compounding leverage rather than bigger budgets, not only for their customers, but for themselves.

Forbes