Maisa has secured $25 million in seed funding led by Creandum, which will be used to expand AI research, engineering, sales, and customer success, particularly in Europe and North America.

The company focuses on developing accountable AI agents that can learn, be supervised, and operate transparently, contrasting with traditional black-box models.

Maisa has launched Maisa Studio, a model-agnostic platform that enables users to deploy trainable digital workers through natural language commands, supporting complex, knowledge-intensive tasks.

Central to Maisa's technology is the Knowledge Processing Unit (KPU), a deterministic reasoning engine that turns large language models into logical, step-by-step executors, reducing hallucinations and increasing transparency.

The platform incorporates a Chain-of-Work, an audit trail that logs every decision and action, ensuring transparency, compliance, and the ability to replay processes, especially in regulated industries.

Maisa's digital workers are being piloted across industries like banking, automotive, and energy to automate complex workflows with full traceability, resulting in significant productivity gains and reduced false positives.

Early implementations include a financial firm reducing false positives by 99% and increasing productivity tenfold, demonstrating the platform’s effectiveness in compliance-heavy environments.

Maisa emphasizes reliability, auditability, and the ability to fix AI systems, aiming to build trustworthy solutions that support regulatory compliance and internal audits while systematically reducing hallucinations.

Despite the high failure rate of AI pilots in enterprises, Maisa’s technology aims to provide practical, trustworthy AI solutions that address reliability and safety concerns.

The startup plans to grow its team from 35 to 65 employees by early 2026 and expand its customer base, especially in regulated sectors, with a presence in Valencia and San Francisco.

Founded in 2024 by AI experts David Villalón and Manuel Romero, Maisa develops model-agnostic AI platforms that transform large language models into trustworthy, specialized AI agents for regulated industries.

Maisa positions itself as an advanced form of robotic process automation (RPA), capable of delivering productivity gains without extensive manual programming or rigid rules.