Top 10 AI Startups to Watch in Andorra in 2026

Table of Contents

• Introduction: Andorra's AI piste map
• Nexad
• PRAAMS
• ZKON Network
• Floatist
• Laennec AI
• Limbit
• Bet Hero
• Yacht Maintenance Program
• TONEO
• RocheGrup Softwares
• Conclusion: Reading the map, riding the mountain
• Frequently Asked Questions

Nexad

Nexad is the name that keeps coming up whenever founders in Andorra la Vella talk about AI with global ambitions. Headquartered in the principality, the company closed a $6 million seed round in April 2025 led by Prosus and Andreessen Horowitz (a16z), a rare combination of investors for such a small market. In Tracxn’s Andorra startup tracker, Nexad consistently appears in the country’s top-funded software ventures.

Problem & product

Digital advertising teams are trapped between performance targets and tightening privacy rules. Nexad’s platform ingests real-time interaction data such as scroll depth, hover behaviour, and dwell time, then uses generative models to assemble and place ad creatives on the fly. Instead of manually defined campaigns, brands get an engine that rewrites copy, adjusts layout, and shifts placements in milliseconds, with a strong focus on contextual understanding rather than third-party cookies.

Why Andorra matters

What makes Nexad interesting for this valley is how it uses Andorra as a secure, cost-efficient base. The company processes data under European-grade privacy expectations while benefiting from the principality’s low corporate tax regime and lean operating costs. Ecosystem surveys like Andorra Startup’s innovation overview point to adtech and fintech as early pillars of the local digital economy; Nexad sits squarely in that first camp, branding itself as “AI-native” infrastructure rather than a bolt-on optimisation tool.

What to watch next

Over the next few seasons, the main questions are about scale and regulation. With tier-one VCs already on the cap table, Nexad is a credible candidate for rapid expansion into North America or the Gulf. At the same time, its contextual targeting model will be tested against evolving EU privacy and AI rules: the more its optimisation remains explainable and cookie-independent, the stronger its position with large European advertisers and regulators alike.

PRAAMS

Fintech is already the best-funded tech vertical in Andorra, with capital concentrating around asset management and adtech. Within that flow, PRAAMS has emerged as a flagship WealthTech player, featured in The Wealth Mosaic’s AI WealthTech market map as part of a new wave of “vertical AI” for portfolio construction.

Problem & product

Traditional portfolio tools are slow, manual, and often restricted to a narrow instrument universe. PRAAMS offers an automated cross-asset research and optimisation engine that can scan over 110,000 stocks and bonds in seconds, then build efficient portfolios around client constraints such as risk tolerance, ESG filters, or liquidity needs.

• Delivered as a white-label API for banks, advisors, and robo-platforms
• Typical integration time of about one week into existing front ends
• Backed by explainable models that surface human-readable rationales

On StartupBlink’s Andorra rankings, PRAAMS holds a score of 61, placing it near the top of local fintech ventures and signalling meaningful traction with regulated institutions.

Why Andorra is a fit

PRAAMS is almost purpose-built for Andorra’s role as a regional wealth hub: high-net-worth residents, cross-border Spanish and French clients, and private banks like Andbank, Morabanc, and Crèdit Andorrà under pressure to differentiate digitally. Operating from a low-tax jurisdiction lets the company reinvest heavily in research and compliance while staying a short drive from Barcelona’s fintech scene and institutional investors around Toulouse. According to OpenVC’s overview of Andorran investors, this cross-border reach is exactly what local capital is beginning to reward.

What to watch

The key milestones now are depth of bank integrations, alignment with MiFID II and the EU AI Act’s explainability requirements, and potential acquisition interest from larger robo-advisors or core banking vendors once PRAAMS reaches a critical mass of assets powered by its optimisation engine.

ZKON Network

As more AI agents and DeFi protocols act autonomously on financial data, the core question shifts from “Can we access data?” to “Can we prove it’s correct?” ZKON Network is built around that trust gap, positioning itself as the cryptographic backbone for AI systems that depend on on-chain and off-chain feeds.

Problem & product

Most oracle solutions today centralise control or rely on weak economic incentives, leaving AI workflows exposed to manipulation. ZKON tackles this with a decentralised oracle network that combines Zero-Knowledge Proofs (ZKPs) and Multi-Party Computation (MPC). Data consumers - from smart contracts to autonomous AI agents - can verify, mathematically, that what they receive is both untampered and privacy-preserving.

• Verifiable data feeds for multiple chains
• Privacy-preserving aggregation of off-chain sources
• Design explicitly tailored to AI and DeFi co-existing

Funding and ecosystem position

Within Andorra’s Web3 cluster, ZKON is in the “most funded” tier of startups. Its broader segment has helped push local Web3 and AI ventures to over $14 million in combined funding, with projects like Blockchain Monster Hunt also contributing meaningful capital. According to TechBehemoths’ overview of AI companies in Andorra, this crypto-AI overlap is one of the country’s fastest-moving niches.

Why Andorra is a strategic base

Andorra offers a pragmatic mix for ZKON: comparatively friendly treatment of digital assets, low corporate taxation, and high-quality fibre connectivity suitable for latency-sensitive oracle infrastructure. At the same time, the team remains plugged into Spanish and French blockchain communities along the trans-Pyrenean corridor, where events and accelerators spotlight projects like ZKON on platforms such as F6S’ list of top Andorran startups.

What to watch

For the next phase, the critical markers are deep integrations with major L1 and L2 ecosystems, early deployments where autonomous AI agents call ZKON to trigger on-chain actions, and potential M&A interest from larger oracle or rollup providers looking to add strong cryptographic guarantees to their stack.

Floatist

From Andorra la Vella it’s only a morning’s drive down the CG-1 to Barcelona’s port, which makes Floatist’s “remote” HQ feel less improbable than it looks on a map. The company is using that vantage point to build a category-defining platform for the Mediterranean yacht charter market, where most operators still juggle bookings, crews, and maintenance across spreadsheets and WhatsApp threads.

Floatist offers a single cloud system that automates charter administration, orchestrates mobile check-in/check-out workflows, and tracks vessel condition to schedule predictive maintenance. Under the hood, its models learn from historical incident reports, engine hours, component replacements, and environmental data to flag high-risk boats before they fail - reducing both downtime and insurance headaches.

• Centralised calendar and contract management for charter fleets
• Mobile apps for crews and clients at embarkation and return
• ML-driven alerts tied to actual usage patterns and service history

That focus on yachting-specific data is why Floatist ranks #1 in StartupBlink’s “Software & Data” category for Andorra, signalling genuine traction in a demanding vertical. The team can validate features quickly with fleets in Barcelona, Palma, and the Côte d’Azur, then return the same day to build from a low-tax Andorran base. Seasonal networking also helps: May alone brings 15+ international tech and AI events to the country, according to AllConferenceAlert’s conference listings for Andorra, giving Floatist regular access to partners and talent without leaving the valley.

For AI and MLOps engineers, this is a textbook vertical AI environment: highly structured operational data, clear business value in reducing unplanned downtime, and clients willing to pay for reliability. As guides like Neowork’s analysis of outsourcing AI engineers point out, such domain-rich problems are where specialised teams can defend margins against generic tooling.

Laennec AI

Healthcare innovation in the Pyrenees looks very different from adtech or crypto. Instead of click-through rates or gas fees, the stakes are diagnostic accuracy, patient safety, and CE-marking. That’s the terrain Laennec AI has chosen: building clinical decision-support tools from an Andorran base where the health system is compact, but the regulatory bar is fully European.

From data overload to decision support

Clinicians are drowning in imaging, lab results, and unstructured notes, while consultation slots get shorter. Laennec AI is developing AI-powered diagnostic assistance that surfaces risk scores, differential diagnoses, and anomaly flags directly inside existing workflows. Rather than forcing doctors into new software, its ambition is to sit quietly behind the scenes, prioritising cases and highlighting patterns that might otherwise be missed.

The company has secured around $400,000 in pre-seed funding via the OpenVC investor network, giving it just enough runway to tackle the hardest early work: data partnerships, clinical validation, and regulatory planning. Analyses such as Interexy’s review of leading telemedicine builders underline how much of global health innovation now hinges on robust, AI-ready data pipelines - exactly the infrastructure Laennec has to negotiate with local hospitals.

Andorra as a micro-testbed

Andorra’s scale is an advantage here. A single national hospital, a mix of local residents and high-altitude sports injuries, and strong ties to Catalan and Occitan health systems create a controllable sandbox for pilots. The government’s broader push to make AI a pillar of its “technological hub in the Pyrenees” means HealthTech experiments get political attention as well as clinical scrutiny.

Globally, initiatives like CIFAR’s work on AI and neurology highlight how decision-support systems are moving from lab curiosity to bedside tool. Laennec’s challenge is to translate that frontier research into CE-marked, explainable products that Andorran clinicians trust enough to use on real patients - and then prove it scales across the border to Barcelona or Toulouse.

Limbit

In Andorra, tourism is the engine that keeps the valleys humming: millions of visitors each year translate into a dense grid of restaurants, hotels, and bars packed into Andorra la Vella, Escaldes-Engordany, and the ski stations. Yet most of those venues still manage staffing, inventory, and menu changes with gut feel and end-of-week spreadsheets. Limbit is building the missing data layer for that hospitality network.

From gut decisions to real-time signals

Limbit offers an all-in-one digital platform for the food and beverage industry that continuously collects quality and operational data in real time - prep times, ticket volumes, waste, stock levels - and runs predictive analytics over it. The goal is simple: give non-technical owners the kind of live dashboards and forecasts usually reserved for big retail chains.

• Demand forecasting by hour, day, and season
• Inventory optimisation to cut waste and stockouts
• Operational KPIs on service speed, consistency, and margins

Why this is a classic SME AI play

Analysts tracking AI for small businesses, such as the team behind AdventurePPC’s overview of practical AI use cases, underline the same pattern Limbit is exploiting: once data collection is automated and low-friction, even small venues can use forecasting and optimisation without in-house data scientists.

For a mid-sized restaurant in Escaldes-Engordany, that might mean predicting ski-weekend demand, adjusting rosters, and ordering just enough fresh produce to ride the rush without throwing half of it away on Monday. Multiply that across dozens of venues in Pas de la Casa or Ordino-Arcalís, and the aggregated patterns become a powerful training signal.

Andorra as a testing ground

Because Andorra’s hospitality market is compact but data-rich, Limbit can pilot features quickly, iterate with local owners, and then export the same playbook to other mountain corridors. Global events like the Google AI Summit keep its team aligned with best practices in operational AI, while the country’s low tax rates make it easier to reinvest subscription revenue into better models and integrations with POS and hotel systems.

Bet Hero

Tucked between Spain and France, Andorra is geographically small but sits on huge sports and betting audiences just down the valleys. Bet Hero uses that vantage point to run a data business rather than a licensed book: a real-time sports betting analytics platform that treats odds the way high-frequency traders treat prices.

Product: odds as live data

Bet Hero’s engine continuously scans 400+ sportsbooks with sub-second updates, calculating where lines are mispriced or out of sync. The platform flags positive expected value (+EV) and arbitrage opportunities, pushing alerts to dashboards tuned for both casual bettors and professional syndicates.

• High-frequency line and odds monitoring across operators
• Automated +EV and arbitrage detection
• Subscription plans starting at €49.99 per month

Why Andorra works for this model

Because Bet Hero sells analytics rather than taking bets, it can base itself in Andorra la Vella and serve customers in Spain, France, and beyond. The low corporate tax band (around 2-10%) leaves more margin to reinvest in infrastructure, while fibre connectivity keeps latency low enough for serious line-shopping. Regionally, AI investment is accelerating; as Wamda’s report on AI capital flows notes, data-intensive verticals like trading and betting are among the first to monetise real-time models.

What to watch from here

The big variables on this run are regulation and commercial execution. EU and national rules on betting data, scraping, and consumer protection could reshape which products Bet Hero can sell and to whom. On the upside, the same infrastructure could power fantasy sports, player-prop modelling, or media-facing odds tickers; getting the pricing and go-to-market right, as frameworks like Alexander Group’s guidance on AI commercial models argues, will be as important as the algorithms themselves.

Yacht Maintenance Program

Long before “vertical AI” became a buzzword in Andorra, Yacht Maintenance Program (YMP) was already wiring boats to the cloud. Founded in 2013 and based in Andorra la Vella, it started as a digital logbook and gradually evolved into an AI-driven safety and reliability platform for global yacht fleets.

From logbooks to predictive maintenance

YMP centralises everything an operator needs to keep vessels safe and compliant: maintenance records, work orders, and regulatory documentation, plus growing streams of sensor and usage data. Those data feeds now flow into predictive models that forecast failures and optimise intervention windows, shifting maintenance from reactive to planned.

• Structured history of component replacements and inspections
• Usage profiles by engine hours, routes, and environmental conditions
• Alerts when risk exceeds predefined thresholds

Industrial players across sectors are moving this way; as Avnet notes in its overview of AI for smarter devices, embedding intelligence close to equipment is becoming standard for high-value assets. YMP’s advantage is a decade of yacht-specific data that newer entrants simply don’t have.

Why Andorra works for a maritime platform

Operating from Andorra gives YMP a neutral, tax-efficient jurisdiction to serve owners and flags from many countries, while still being a short trip from major Mediterranean hubs like Barcelona, Palma, or Nice. Locally, it appears among the leading software ventures on Andorra-focused startup maps, and benefits from proximity to newer maritime-AI neighbours such as Floatist, with whom deeper integrations could yield a near end-to-end charter-and-maintenance stack.

The steady blue run

Looking ahead, the key questions are how aggressively YMP doubles down on AI versus remaining “just” a digital logbook, whether it can turn its dataset into an industry reference via OEM and insurer partnerships, and how it positions itself if marine ERP vendors or IoT giants come hunting for acquisitions in the sector.

TONEO

While most Andorran AI ventures chase finance, logistics, or betting, TONEO (often branded as ToneoColor) has picked a different line down the mountain: applying computer vision and deep colour science to the global imaging and digital video industry. Its customers aren’t bankers or fleet managers, but colourists, editors, and content creators who care obsessively about how each frame looks on a viewer’s screen.

From raw pixels to controlled perception

Creators today wrestle with inconsistent colour across cameras, displays, and platforms. TONEO develops hardware and software tools that analyse image and video content frame by frame, model human visual perception, and then optimise grading and contrast for specific output devices. Instead of relying on presets, teams can dial in precise responses for HDR TVs, mobile screens, or cinema projectors, keeping a consistent “look” across campaigns and channels.

As AI-driven discovery reshapes how content is surfaced, that control matters. Guides like Yotpo’s 2026 content gap analysis for AI search underline that brands now compete on how well their assets are tuned for both algorithms and human attention; TONEO’s tools sit exactly at that intersection for visual material.

Why base a colour lab in Andorra?

Running from Andorra lets TONEO play a global niche game from a low-tax, high-quality-of-life microstate, while staying close to Barcelona’s film, gaming, and advertising studios for co-creation with demanding early adopters. The surrounding Pyrenean landscape also doubles as a live test set: high-contrast snow scenes, fast-changing mountain light, and vivid resort branding are ideal for stress-testing grading algorithms.

Looking ahead, the most interesting questions are about integration and reach. Seamless plugins into dominant editing suites, potential bundling with monitors or grading panels, and the shift from rules-based correction to generative visual enhancement will determine whether TONEO remains a specialist tool or becomes foundational infrastructure in the creator economy. Global programmes such as the AI for Good Innovation Factory illustrate how fast AI-powered media tooling is professionalising; TONEO aims to give Andorra a visible place in that story.

RocheGrup Softwares

Not every important AI company in Andorra shows up as a pure SaaS product. RocheGrup Softwares, headquartered in Ordino, is a services-first player that quietly powers other people’s launches. With a team of 10-49 specialists, it focuses on custom builds across generative AI, LLM integrations, and Web3 backends, acting as a technical co-founder for corporates that move slower than startups but control bigger budgets.

From brief to deployed product

RocheGrup’s projects typically run end to end: discovery workshops, architecture, model selection or fine-tuning, frontend and backend development, and deployment. One month it might be a multilingual chatbot over a bank’s knowledge base; the next, a tokenised loyalty scheme for a ski resort or a data layer to make contact centres “agentic.” As enterprise examples like Salesforce’s AI-driven contact centre show, demand is shifting from raw models to deeply integrated workflows - exactly the kind of work boutiques like RocheGrup are set up to deliver.

Why incumbents choose a local builder

Andorran banks, Andorra Telecom, and major hospitality operators know they need AI and Web3 capabilities but struggle to hire full in-house teams. RocheGrup offers a culturally fluent, near-shore alternative to big-city consultancies: close enough for regular workshops, but priced competitively thanks to low personal and corporate tax rates and Andorra’s lower cost base. At a time when, as Futurum Group’s coverage of recent mega-rounds highlights, capital is pouring into AI infrastructure, there is parallel demand for specialists who can turn that infrastructure into working products.

From agency to product studio

The real upside for RocheGrup lies in productising repeatable components they build again and again: KYC pipelines, document-understanding engines for banks, or reusable Web3 modules. Doing so would shift them from pure services to a hybrid agency-product studio, creating recurring revenue and, potentially, spin-out startups. As more AI and Web3 talent relocates to the Pyrenees for lifestyle and tax reasons, RocheGrup could become one of the nuclei where Andorra’s next generation of vertical AI products is quietly assembled.

Conclusion: Reading the map, riding the mountain

Step back from the individual stories and the pattern comes into focus: Andorra is quietly becoming a laboratory for vertical AI. Instead of chasing a generic “AI hub” narrative, the country’s founders are carving precise lines through wealth management, maritime operations, hospitality, betting, media tools, and Web3 infrastructure. Each startup on this list is less a dot on a map than a different face of the same mountain.

For investors used to billion-euro ecosystems, the principality can feel almost too small to matter. Yet the same compactness that keeps Andorra off mainstream rankings is what makes it an efficient testbed. Talent, regulators, and enterprise buyers sit close enough for quick iterations, while the trans-Pyrenean corridor to Barcelona and Toulouse opens into far larger markets. In a European context where, as one investor warned in The Quantum Insider’s analysis of the AI race, the continent risks falling behind on core platforms, these tightly focused, revenue-driven experiments matter.

For AI and ML professionals living in or eyeing Andorra, the implication is practical: don’t just ask “Which is the best startup?” Ask instead which run you want to drop into. A WealthTech engine like PRAAMS demands different skills and risk tolerance than a Web3 oracle such as ZKON, a maritime MLOps platform like Floatist, or a colour-science specialist such as TONEO. Each line offers its own mix of domain expertise, regulatory exposure, and exit paths.

Treat this Top 10 as a piste map, not gospel. Use it to pick a starting gate - an intro email, a diligence call, a coffee in Andorra la Vella before someone drives down to Barcelona for meetings. Then look beyond the board: to bootstrapped teams not yet tracked by databases, to innovation units inside banks and Andorra Telecom, to students and founders pitching in regional showcases like the “Smart Business Story” spotlight on emerging AI startups. The real terrain only reveals itself once you’re moving.

Frequently Asked Questions