Saturday, March 14, 2026

Maisa Hits 400% Growth Building What 95% of Enterprise AI Projects Are Missing: Accountability

By , in AI Spain on . Tagged width: , , , , , , , , , , , , ,

The Valencia startup just earned Deloitte’s Rising Star recognition after growing its client base fivefold in a year.

Its pitch: enterprises don’t need more AI pilots — they need auditable digital workers that can survive regulatory scrutiny. Backed by $25M from Creandum and Forgepoint, Maisa is betting that traceability beats probabilistic magic.

Maisa, an enterprise AI infrastructure company building auditable digital workers for regulated industries, has reported 400% year-on-year growth and a fivefold increase in its client base over the past year.

MAISA 2026 Deloitte's Rising Star
MAISA 2026 Deloitte’s Rising Star

Deloitte has named the Valencia-based startup a Rising Star in its Technology Fast 50 Programme 2026, recognising high-growth technology companies with strong scalability and market momentum. The timing is not coincidental. As enterprises move generative AI beyond experimentation and into mission-critical production environments, the gap between impressive pilots and deployable systems has become painfully visible.

Industry research shows that while nearly 88% of organisations now use AI in at least one business function, only about 38% have successfully scaled AI beyond pilot projects into full production. The bottleneck is not capability — it is governance. Roughly 95% of enterprise AI pilots fail to deliver meaningful ROI without robust infrastructure and control mechanisms. When scrutiny from regulators and internal risk teams increases, enterprises are prioritising infrastructure that combines AI capability with traceability and accountability. Maisa’s growth reflects that shift.

Why 95% of AI Pilots Fail

“Innovation alone doesn’t move AI into production, but accountability does,” said David Villalón, co-founder and CEO of Maisa. “Enterprises are no longer satisfied with impressive pilots. They need systems they can inspect, govern, and defend under audit. Our growth reflects that shift in market expectations.”

Villalón knows this problem firsthand. He studied computer engineering and business administration at Universitat Politècnica de València and launched his first startup, Walnov, while still in university. He later joined Voicemod’s core team as Director of Product during its hyper-growth phase, scaling the company from 14 to 200 people. At Voicemod, he experimented deeply with early large language models like GPT-2 and GPT-3, spotting their limitations — including hallucinations — well before they became mainstream concerns. After Voicemod, he served as Chief AI Officer at Clibrain, leading real-world LLM projects that revealed the enterprise gaps in AI reliability that most vendors were ignoring.

That experience led to Maisa. Rather than building another layer of abstraction on top of probabilistic models, Villalón and co-founder Manuel Romero built infrastructure that structures AI execution into verifiable, step-by-step workflows. The result is what the company calls auditable digital workers — AI systems that execute complex, end-to-end business processes with full transparency. Every action is recorded in a verifiable Chain of Work, allowing enterprises to see what was done, why it was done, and how outcomes were reached. For regulated industries, that distinction is not a nice-to-have feature. It is the baseline requirement for production deployment.

How Maisa’s Architecture Actually Works

Maisa’s platform is powered by its proprietary Knowledge Processing Unit (KPU), which combines contextual AI reasoning with deterministic execution logic and built-in validation at each step. This is not a wrapper around a foundation model. It is an execution layer designed to reduce operational risk, prevent unverified outputs from propagating through workflows, and ensure results remain auditable under regulatory review. The architecture reflects a core insight: in regulated industries, probabilistic magic is not enough. Enterprises need deterministic control.

That architecture is proving its value in early enterprise deployments. Elecnor, a global engineering and infrastructure group, recently implemented Maisa to operationalise internal processes through AI-driven execution.

The Director of Digital Transformation and Innovation at Elecnor described the deployment: “We have been very happy about Maisa’s adaptation to our needs and helping our team succeed with operationalising our standard operating procedures. It has been a collaborative experience where, within a matter of weeks, we moved from semi-structured operational guidance to a fully functional Digital Worker that enables us to manage our first AI-driven business operation, while simultaneously formalising a written standard operating procedure aligned with our business goals.”

That speed matters. Moving from operational guidance to production deployment in weeks — not quarters — is the competitive advantage enterprises are looking for. Other deployments include stabilising a car manufacturer’s supply chain during a crisis and automating risk assessments for financial services firms. These are not edge cases. They are exactly the kind of documentation-heavy, judgment-based workflows that have historically resisted automation because traditional systems could not deliver the transparency compliance teams demand.

The Funding and the Roadmap

Maisa raised a $5 million pre-seed round in December 2024 from NFX and Village Global, followed by a $25 million seed round in August 2025 led by Creandum and Forgepoint Capital International, with continued participation from NFX and Village Global. That $30 million in total funding is being deployed toward product development and enterprise expansion as the company deepens its presence in Europe and the United States. Over the past year, Maisa has grown its team fivefold while expanding deployments in highly regulated sectors, including finance, engineering, manufacturing, and infrastructure.

The investor thesis is straightforward: as AI adoption matures, infrastructure capable of delivering transparent, traceable, and defensible outcomes is becoming foundational to enterprise deployment. Creandum and Forgepoint are both known for backing infrastructure-layer companies that solve structural problems, not feature gaps. Their participation signals conviction that accountability infrastructure — not model performance — is the bottleneck preventing AI from scaling into regulated production environments.

Why This Matters Now

Villalón has been vocal about why enterprise AI deployments fail. In talks at VDS 2025 and Scaling Europe, he has emphasised that 95% of AI pilots fail not because the models are insufficient, but because governance, integration complexity, and limited visibility into AI-driven decision-making prevent scale. Enterprises experimenting with generative AI quickly discover that probabilistic outputs are not enough when legal, compliance, and risk teams demand explanations. Without infrastructure capable of producing verifiable audit trails, AI remains confined to low-stakes use cases.

Deloitte’s Rising Star recognition underscores a broader market shift: the AI infrastructure layer is no longer about maximising model performance — it is about making AI deployable under real-world constraints. Maisa’s 400% growth and fivefold client expansion suggest that enterprises are ready to move past pilots and into production, but only if they can defend those systems under audit. For a Valencia-based startup competing against well-funded incumbents, that is not just a market opportunity. It is the only defensible position in a category where trust, not capability, is the constraint.

Deloitte's Rising Star Celebration 2026 Fast50
Deloitte’s Rising Star Celebration 2026 Fast50

More Deep Tech Startup Coverage

More Quantum Computing Coverage