A recent report published by MIT’s Nanda Initiative shows that 95% of corporate generation AI pilots have failed. However, the most advanced organizations are experimenting with agent AI systems that can learn and monitor, rather than abandoning technology altogether.
That’s where Maisa AI appears. A year ago, the startup built an overall approach around the premise that enterprise automation requires an accountable AI agent rather than an opaque black box. The new $25 million seed round, led by European VC company Creandum, has now launched Maisa Studio, a model-independent self-service platform that helps deploy digital workers who can train in natural languages.
It may sound familiar, but when you recall adorable, so-called vibe coding platforms with cursors and cremated back, Mysa argues that the approach is fundamentally different. “Instead of using AI to build responses, we use AI to build the process that we need to run to do what we call “workplace,” Maisa CEO David Villalón told TechCrunch.
The main architect behind this process is Manuel Romero, co-founder and chief science officer of Maisa, who previously worked with Villaron at Spanish AI startup Cribrain. In 2024, the duo worked together to build a hallucination solution after seeing firsthand that they “we couldn’t rely on AI.”
The pair are not skeptical of AI, but they believe that it is not feasible for humans to review “three months of work done in five minutes.” To address this, Maisa employs a system called HALP (Human Advanced LLM Processing). This custom method works like a student on Blackboard. This asks users what their needs while they outline each step that a digital worker will follow.
The startup has also developed a Knowledge Processing Unit (KPU), a deterministic system designed to limit hallucinations. Maisa started with this technical challenge rather than a use case, but soon found it resonated with companies that bet on reliability and accountability and want to apply AI to critical tasks. For example, clients currently using Maisa in production include large banks and companies in the automotive manufacturing and energy sector.
By serving these enterprise clients, Maisa hopes that companies will position themselves as a more advanced robotic process automation (RPA) that unlocks productivity gains without relying on strict predefined rules or extensive manual programming. To meet their needs, startups provide deployments through secure cloud or on-premises deployments.
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
This enterprise-first approach means that Maisa’s customer base is still very small compared to the millions who flock to code platforms with a freemium atmosphere. However, as these platforms are currently exploring ways to attract enterprise customers, Maisa is moving in the opposite direction in Maisa Studio. It is designed to increase customer funnels and facilitate recruitment.
The startup is also planning to expand with existing customers operating in multiple countries. With double headquarters in Valencia and San Francisco, Maisa has already posted scaffolding in the US, which is reflected in its cap table. The $5 million seed round was led by San Francisco-based ventures NFX and Village Global.
Additionally, TechCrunch has exclusively learned that US Forgepoint Capital International has joined the new round via a European joint venture with Spanish bank Banco Santander, highlighting its appeal to the regulatory sector.
Focusing on complex use cases that demand accountability from non-technical users can be a differentiator for Maisa. Maisa’s competitors include business-focused workflow automation products powered by Crewai and many other AI. In a LinkedIn post, Villalón highlighted this “AI Framework Gold Rush” and warned that “quickstart” will be a long nightmare if you need reliability, auditability, or the ability to fix something wrong. ”
Double its goal of supporting AI scales, Maisa plans to use its funds to grow from 35 to 65 by the first quarter of 2026 to meet demand. Starting from the last quarter of the year, startups expect rapid growth as they begin offering waitlists. “We show the market that there is a company offering what is promised and that it is working,” Villaron said.
