A New Era in AI: Maisa AI Addresses the 95% Failure Rate of Generative AI Pilots

A staggering 95% of generative AI pilots at companies are failing, according to a recent report published by MIT’s NANDA initiative. However, forward-thinking organizations are not giving up. Instead, they are exploring agentic AI systems capable of learning and being efficiently supervised.

Introducing Maisa AI: Pioneering Accountable AI Solutions

This is where Maisa AI steps in. The innovative startup has built its foundation on the belief that enterprise automation needs accountable AI agents rather than obscure black boxes. With a recent $25 million seed funding round led by European VC firm Creandum, Maisa has unveiled Maisa Studio—a model-agnostic self-service platform that enables users to train digital workers using natural language.

A Unique Approach to AI-Driven Processes

While the concept might seem akin to vibe coding platforms like Cursor and Lovable, Maisa emphasizes a fundamentally different methodology. “Instead of using AI to generate responses, we leverage AI to construct the processes required to achieve those responses—what we call ‘chain-of-work,’” says CEO David Villalón.

The Visionary Team Behind Maisa AI

Leading this innovative approach is co-founder and Chief Scientific Officer Manuel Romero, a former colleague of Villalón at the Spanish AI startup Clibrain. They recognized the limitations of AI in 2024, expressing the need for a solution that mitigated hallucinations, understanding that “you cannot solely rely on AI,” as Villalón articulated.

HALP: Human-Augmented LLM Processing

Maisa introduces HALP, or Human-Augmented LLM Processing, a unique system that encourages user interaction while digital workers delineate their operational steps. This approach resembles students solving problems at a blackboard, making the process more collaborative.

Building Trust with the Knowledge Processing Unit

Maisa also developed the Knowledge Processing Unit (KPU), a deterministic mechanism designed to curb hallucinations. Initially focused on solving technical challenges rather than specific use cases, Maisa soon discovered that its emphasis on reliability resonated with organizations eager to apply AI responsibly—ranging from a prominent bank to firms in the automotive and energy sectors.

Revolutionizing Robotic Process Automation

By catering to enterprise clients, Maisa aims to redefine robotic process automation (RPA), facilitating productivity boosts without the need for rigid, predefined protocols or extensive manual coding. The startup offers deployment in both secure cloud and on-premises environments to meet diverse organizational needs.

Strategizing for Future Growth in the AI Landscape

Despite its enterprise-first orientation resulting in a smaller customer base compared to the millions flocking to freemium vibe-coding platforms, Maisa aims to capture market share as these competitors pivot to enterprise appeal. The launch of Maisa Studio is designed to expand its customer funnel and facilitate adoption.

Global Expansion Plans and Industry Partnerships

Maisa plans to scale with existing clients with operations across various countries. With headquarters in Valencia and San Francisco, Maisa is set to strengthen its foothold in the U.S.; its $5 million pre-seed funding last December was led by San Francisco venture firms NFX and Village Global.

Attracting Investment for Regulated Sectors

TechCrunch has learned that U.S. firm Forgepoint Capital International participated in this latest funding round through its European venture with Banco Santander, emphasizing Maisa’s appeal within regulated industries.

Maisa’s Unique Position in the Competitive AI Marketplace

Focusing on intricate use cases that demand accountability from non-technical users could set Maisa apart from competitors like CrewAI and various other AI-driven workflow automation tools. In a recent LinkedIn post, Villalón underscored the pitfalls of the “AI framework gold rush,” warning that shortcuts can lead to long-term complications when reliability, auditability, and corrective measures are needed.

Doubling Staff to Meet Demand and Deliver Results

With ambitions of expanding its team from 35 to 65 by the first quarter of 2026, Maisa is poised for rapid growth. Starting in late 2025, the startup expects to serve its waiting list and prove that it can deliver on its promises. “We are going to show the market that there is a company that is delivering what has been promised, and that it’s working,” Villalón asserts.