Barcelona-based startup Mafer AI, founded by entrepreneurs Fer Oliver Jané and Marc Montalbo Burges, has closed a €2 million pre-seed funding round led by Kfund, 4Founders Capital, Masia, and Lavanda Ventures — the startup investment vehicle backed by the Puig family.
The round also attracted support from prominent business angels and international technology operators, including Adrián Mato (Andreessen Horowitz scout and GitHub Copilot executive), Manuel Roca (founder of Atrápalo and investor in Wikiloc), Dídac Lee, Fernando Castiñeiras (Managing Partner at Athos Capital), and Juanjo Mostazo. The investor base additionally includes professionals from organizations and companies such as MIT, Google, Samaipata, Bain, BCG, Shakers, and Happy Robot, as well as several investors specialized in the cosmetics and fragrance industries.
A New AI Infrastructure for Product Formulation Industries
Mafer AI develops MaferOS, an artificial intelligence platform designed to help R&D teams across industries such as chemicals, food and beverage, cosmetics, fragrances, personal care, pharmaceuticals, and nutraceuticals structure, analyze, and leverage decades of accumulated technical knowledge.
The company is addressing a structural yet largely invisible bottleneck within formulation-driven industries: the inability to effectively utilize the vast amount of technical historical data generated across laboratories and formulation processes. Failed formulas, analyses, regulatory documentation, technical decisions, and experimental data remain fragmented across spreadsheets, PDFs, proprietary instrument formats, and, most importantly, within the tacit knowledge of senior experts.
This fragmentation slows innovation cycles by five to ten times compared to the competitive pace of the market, erodes margins, and delays product launches. In many cases, when key senior experts leave a company, years of institutional knowledge disappear with them.
Proprietary Models Trained on Each Client’s Data
Mafer AI’s approach combines advanced artificial intelligence models with a proprietary architecture tailored to each customer. The platform trains proprietary models exclusively on each company’s historical data while ensuring full data privacy and protection.
MaferOS is structured through specialized modules capable of automating processes that are currently performed manually or through fragmented workflows, including:
Rather than simply applying AI to chemistry or formulation, Mafer AI is introducing into this sector the next generation of enterprise software architecture already reshaping other industries: proprietary client-trained models, specialized AI agents operating on structured data layers, and embedded Forward Deployed Engineers working directly within organizations to bring products into production within weeks.
This operational model, popularized by companies such as Palantir and recently validated by enterprise players like Happy Robot and Wonderful, is being applied for the first time to the specialty chemicals and FMCG industries.
Growth, Clients, and Technological Expansion
Just six months after its founding, Mafer AI has already secured nearly €300,000 in committed ARR and expects to reach €1 million ARR by the end of 2026.
The company currently works with three production clients in the fragrance and flavor vertical, where it has already structured more than 50,000 technical documents. In parallel, Mafer AI maintains an active enterprise pipeline across six different industry verticals and is running proof-of-concept projects in four of them. The company has also won three competitive procurement processes against international consulting firms.
The team, currently composed of eight people — including experts in chemistry, mathematics, physics, biomedical engineering, and data engineering — plans to grow to twelve professionals by the second quarter of 2026.
Backed by the Barcelona Supercomputing Center
As part of its technological roadmap, Mafer AI has been selected by the Barcelona Supercomputing Center for both Batch 1 and Batch 2 of its AI Factory program. Through the initiative, the company will gain access to more than 50,000 GPU computing hours on NVIDIA H100 infrastructure and to the MareNostrum V supercomputer, a key asset for advancing the development of its first domain-specific foundational models.
The newly raised capital will primarily be allocated toward expanding the company’s technical team, strengthening its research and Forward Deployed Engineering divisions, deepening deployments with existing customers, and accelerating the roadmap for its proprietary AI models.
“Product formulation industries have spent decades accumulating a silent asset — their historical R&D knowledge — without any scalable way to exploit it. The current generation of AI completely changes that equation. Today, using the data of a single client, we can train proprietary models that dramatically accelerate the speed at which companies bring new formulations to market,” said Fernando Oliver Jané, CEO and co-founder of Mafer AI.
“We are building a new technology category from Europe during what we believe is a unique and limited window of opportunity. With a world-class team, the support of the Barcelona Supercomputing Center, and this €2 million financing round, we intend to lead that transformation,” he added.
cloud technology axon
Xcalibur Smart Mapping, recognised as the global leader in airborne...
