Maisa AI raises $25M to tackle enterprise AI reliability issues

A recent MIT report reveals that 95% of generative AI pilots in enterprises fail, prompting the need for more reliable solutions. Enter Maisa AI, a startup focused on creating accountable AI agents rather than opaque systems. The company has secured a $25 million seed round led by European VC firm Creandum to launch Maisa Studio, a model-agnostic platform for deploying trainable digital workers.

How Maisa AI Differs

Unlike traditional AI tools that generate responses directly, Maisa AI builds "chain-of-work" processes to ensure transparency. CEO David Villalón explains: "We use AI to build the process that needs to be executed to get to the response." The startup’s co-founder and Chief Scientific Officer, Manuel Romero, previously worked with Villalón at Spanish AI startup Clibrain.

Key Innovations

• HALP (Human-Augmented LLM Processing): A system where AI outlines each step for user approval, akin to students at a blackboard.
• Knowledge Processing Unit (KPU): A deterministic system to minimize hallucinations, critical for sectors like banking and manufacturing.

Enterprise Focus

Maisa targets regulated industries, offering:

• Secure cloud or on-premise deployment
• Solutions for banks, car manufacturers, and energy companies

Growth Plans

With dual HQs in Valencia and San Francisco, Maisa plans to expand its team from 35 to 65 by early 2026. The startup’s $5M pre-seed round was led by NFX and Village Global, with Forgepoint Capital International joining the latest round.

Competitive Landscape

Maisa competes with platforms like CrewAI but emphasizes reliability and auditability. Villalón warns of the "AI framework gold rush," stressing the need for systems that can be fixed when errors occur.

"We are going to show the market that there is a company delivering what has been promised," Villalón said.