Galtea Raises €2.7M to Build Safer AI Agents

Barcelona's Galtea Secures €2.7 Million in Seed Funding, Backed by Mozilla Ventures, to Revolutionise AI Evaluation

In one of the most notable pieces of AI funding news to come out of Southern Europe this quarter, Barcelona-based startup Galtea has officially closed a €2.7 million (approximately $3.2 million) seed funding round to push forward its ambitious mission of making enterprise AI deployments safer, more reliable, and genuinely production-ready. The round was led by Munich-based early-stage technology fund 42CAP, with notable participation from Mozilla Ventures — the corporate venture capital arm of the Firefox browser developer — alongside existing backers JME Ventures, Masia, and ABAC Nest Ventures. This latest injection of capital brings Galtea's total funding to approximately $4.1 million, following a successful $870,000 pre-seed round secured back in 2024, underscoring the growing investor confidence in the AI testing and evaluation space.

The announcement has sparked significant interest across the European tech ecosystem, and for good reason. As AI agents rapidly move from experimental prototypes to deeply embedded enterprise tools, the infrastructure required to test, evaluate, and validate their behaviour before they touch real users remains frustratingly underdeveloped. Galtea is directly addressing that gap — and the backing from both a forward-thinking VC like 42CAP and a mission-driven investor like Mozilla Ventures signals that the market is finally beginning to take AI quality assurance as seriously as AI development itself. For anyone tracking AI funding news across Europe, this deal represents more than just capital changing hands — it marks a broader structural shift in how the industry thinks about the lifecycle of AI products.

The Problem Galtea Is Solving: When AI Demos Don't Match Real-World Behaviour

There is a persistent and costly disconnect in how AI products are built and how they eventually behave when deployed into production environments. Traditional quality assurance, the kind that has worked reliably for software for decades, was designed to check surface-level functionality. It tells you whether a button renders correctly, whether a user flow executes without errors, and whether an interface behaves as the product team intended. What it cannot tell you is whether the intelligence driving that product — the large language model working silently in the background — is doing what it is supposed to do, responding in alignment with business policy, staying within defined ethical boundaries, and avoiding harmful outputs like hallucinations, bias, and security vulnerabilities.

This is precisely the challenge that Galtea's co-founder and CEO Jorge Palomar identified during his time at the Barcelona Supercomputing Center (BSC), one of Europe's most respected high-performance computing and AI research institutions. Palomar, who previously worked at Amazon before joining BSC, has built his professional career around understanding data governance and the practical challenges of putting AI to work in large organisations. His co-founder and CTO, Dr. Baybars Külebi, holds a doctorate in Machine Learning and brings over a decade of hands-on AI research to the table. The two met while working together at BSC and identified, through repeated practical exposure, that the testing data available to evaluate LLM-based products was simply not realistic enough to catch the kinds of failures that matter most when AI goes live.

Galtea was officially spun out of the Barcelona Supercomputing Center in October 2024, making it one of the more recent additions to Spain's rapidly expanding deep tech startup landscape. But despite being barely eighteen months old, the company has already built a product sophisticated enough to attract marquee enterprise clients like Telefónica, one of the world's largest telecommunications companies, and Abanca, a leading Spanish banking institution — both of which are deploying AI in contexts where errors carry significant consequences.

How the Platform Works: Generating Realistic AI Testing Scenarios

At the heart of Galtea's product is a deceptively simple but technically deep idea: if you want to know whether your AI agent will behave correctly in the real world, you need to test it against scenarios that actually reflect the real world. Most existing testing frameworks rely on static, pre-defined test cases that are either written by hand or pulled from historical data. The problem is that real users are unpredictable. They ask strange questions, use ambiguous language, attempt to push boundaries, and in high-stakes sectors like banking and healthcare, they sometimes ask for things the AI should absolutely not provide.

Galtea's platform addresses this by using AI itself to generate highly realistic, context-specific testing scenarios that simulate the range of situations an enterprise AI agent might encounter. These synthetic yet realistic simulations are designed to probe the AI agent across four critical dimensions: hallucination (generating false or fabricated information), bias (producing outputs that unfairly favour or disadvantage certain groups), security vulnerabilities (susceptibility to prompt injection and other adversarial attacks), and toxicity (generating harmful, offensive, or inappropriate content). The platform integrates directly with popular developer tools and frameworks including GitHub, LangChain, and MLflow, which means it fits naturally into the existing workflows of engineering teams without requiring major process overhauls.

This developer-first approach is a deliberate strategic choice. Rather than positioning the platform as a top-down compliance exercise enforced by legal or risk teams, Galtea has built its tooling to be something developers themselves want to use because it makes their lives easier and their products more robust. It gives individual engineers the ability to run rigorous evaluations continuously throughout the development process rather than treating testing as a final-stage checkbox. For enterprises operating in regulated industries, this shift from post-production auditing to embedded, lifecycle-level governance is not just a nice-to-have — it is rapidly becoming a legal necessity.

Why Mozilla Ventures Got Involved: Trustworthy AI as a Core Investment Thesis

Among the investors in this round, Mozilla Ventures stands out as particularly significant, not just for the credibility its name brings to the table, but for the clarity of the signal it sends about where Galtea is positioned within the broader AI ecosystem. Mozilla Ventures was established in 2022 with an explicit investment thesis centred on supporting companies that are building toward a more trustworthy, human-centric internet. Its portfolio reflects a consistent focus on AI safety, privacy-preserving technology, and digital rights — values that align directly with what Galtea is building.

Palomar has been candid about what Mozilla's participation means beyond the capital itself. Having Mozilla Ventures in the cap table is a strategic signal that Galtea is developing technology in the right direction, actively contributing to a safer and more accountable digital ecosystem. Mozilla also brings something that pure financial investors rarely can: community. Its global network of developers, open-source contributors, and technologists who care deeply about the ethics of the web represents a meaningful distribution and advocacy channel for a company whose product is fundamentally about making AI more trustworthy. This kind of strategic alignment between investor mission and company mission is increasingly rare in today's AI funding landscape, and it has not gone unnoticed by the broader ecosystem of enterprise buyers evaluating which AI governance tools to bring into their organisations.

The involvement of 42CAP, meanwhile, brings a complementary set of strengths. As a Munich-based early-stage fund with a strong track record of backing European deep tech companies, 42CAP provides the kind of grounded, operationally experienced support that a company at Galtea's stage needs as it moves from promising startup to commercially viable enterprise software business. The combination of 42CAP's operational network and Mozilla Ventures' community-driven reach creates a genuinely differentiated support system for Galtea's next phase of growth.

EU AI Act Compliance: A Tailwind Driving Urgent Demand

One of the most important contextual factors behind this round of AI funding is the rapidly evolving regulatory environment in Europe. The EU AI Act, which represents the world's most comprehensive legislative framework for artificial intelligence, has created an entirely new category of compliance obligation for companies deploying AI in high-risk sectors. Under the Act, organisations using AI systems in areas like healthcare, education, financial services, insurance, and public administration are required to demonstrate that their systems have been rigorously evaluated, that risks have been identified and mitigated, and that governance processes are in place to ensure ongoing accountability.

For many enterprises, this is a genuinely difficult challenge. They have invested heavily in building AI products, but they lack the tooling to produce the kind of systematic, documented evidence of evaluation that regulators now require. Galtea's platform sits precisely in this gap. It does not just help teams build better AI — it helps them produce a defensible, auditable record of how their AI was tested, what risks were identified, and how those risks were addressed. This positions the company not as a compliance burden but as an accelerant, helping legal and development teams move faster by providing a structured, integrated approach to governance rather than leaving teams to build their own frameworks from scratch.

This regulatory tailwind is particularly relevant in the sectors Galtea has explicitly chosen to target. Rather than pursuing the broader ecommerce and customer support market, where AI errors are relatively low-stakes and can often be corrected through iterative product improvements, the company is deliberately concentrating on domains where failures carry serious consequences: financial advice that leads to wrong investment decisions, healthcare guidance that could put patients at risk, public administration processes that determine access to government services, and educational platforms that shape how students learn and develop. These are precisely the contexts where the EU AI Act's requirements are most stringent and where the consequences of inadequate testing are most severe.

What the New Capital Will Fund: Engineering, Commercial Expansion, and a New Self-Service Product

The €2.7 million raised in this seed round will be deployed across three primary areas. First, Galtea plans to significantly expand its engineering team, accelerating the development of the platform and adding capabilities that respond to the increasingly complex demands of enterprise AI deployments. Second, the company will invest in building out its commercial function, hiring go-to-market talent and expanding its sales and customer success capabilities as it moves beyond its current base of Spanish and Portuguese enterprise clients.

Third — and perhaps most significant from a product strategy perspective — the funding announcement coincides with the launch of a brand new self-service product offering. Previously, Galtea's platform was available primarily to enterprise clients through direct sales engagements. The new self-service tier allows individual developers and smaller teams to access the platform directly, without the need for a formal sales process. This move dramatically expands the addressable market by lowering the barrier to entry and makes Galtea's tools available to the long tail of development teams building AI products who simply cannot afford or justify an enterprise contract but still need systematic evaluation infrastructure. This is a smart strategic move that mirrors the product-led growth playbooks that have driven the success of tools like GitHub, Stripe, and Datadog — all of which started by winning the hearts of individual developers before scaling into enterprise accounts.

On the geographic front, Galtea's immediate growth ambitions are focused on the UK. While the company has established itself as the go-to AI evaluation tool in the Iberian market — with clients in both Spain and Portugal — it sees the UK as a natural next step given the country's concentration of enterprise technology buyers and its active AI governance regulatory environment. The team has already secured its first small UK customers, and the goal for the coming months is to land significant enterprise logos in London and expand from there into other Northern and Western European markets.

Galtea's Place in the Broader AI Funding Ecosystem

Stepping back, Galtea's seed round is part of a much larger wave of AI funding directed at the infrastructure layer of the artificial intelligence stack. The past eighteen months have seen a dramatic shift in where investor capital is flowing within the AI sector. The first wave of AI funding was heavily concentrated on foundation model development — the expensive, compute-intensive work of training the large language models that underpin almost all AI applications. The second wave focused on application-layer companies building products on top of those models. Now, a third wave is emerging: funding directed at the reliability, safety, governance, and evaluation infrastructure that makes it possible to actually trust and deploy those AI applications at scale.

Galtea is one of the clearest examples of this third-wave dynamic. It is not building another AI model or another AI-powered application. It is building the testing and evaluation infrastructure that makes every other AI product safer and more deployable. In that sense, its position in the market is analogous to what companies like Veracode and Checkmarx did for software security, or what New Relic and Datadog did for application performance monitoring — creating the observability and assurance layer that the broader ecosystem depends on. For enterprises tracking the AI funding news landscape and trying to understand where the next wave of valuable companies will come from, Galtea is exactly the kind of company worth watching closely.

The Barcelona Supercomputing Center spin-off story is also worth highlighting for what it says about the maturation of Europe's AI ecosystem. BSC has long been one of the continent's most respected AI research institutions, but historically, the translation of its research into commercially viable products has been slow and inconsistent. Galtea represents a newer generation of BSC spin-offs that are moving faster, building market-ready products from day one, and attracting international institutional capital — all while remaining rooted in a genuinely rigorous scientific methodology rather than the marketing-forward approach that has characterised too many AI startups.

As AI continues to reshape industries from banking to healthcare to education, the work that companies like Galtea are doing becomes not just commercially important but socially necessary. The stakes of getting AI wrong in high-risk sectors are simply too high to leave evaluation to chance. And with €2.7 million in fresh capital, a world-class founding team, and a growing roster of enterprise clients, Galtea is building something that the AI industry genuinely needs.