Maisa AI Lands $25M to Solve Enterprise AI's Trust Issues

08/28/2025 AI

It seems like every other day I'm reading about another company diving headfirst into generative AI, only to stumble and fall. According to a recent report, a shocking 95% of generative AI projects within companies are failing. That's a pretty grim statistic, right? But instead of throwing in the towel on AI altogether, some organizations are trying a smarter approach: agentic AI systems. These systems can actually learn and be supervised.

That's where Maisa AI comes into play. This young startup is built on the idea that enterprise automation needs AI agents that you can trust, not these mysterious black boxes that no one understands. They just landed a cool $25 million in funding, led by Creandum, a European VC firm, and they've launched Maisa Studio. It's a platform that lets you deploy digital workers that can be trained using plain, everyday language. No need to be a programmer or use complicated code.

Maisa's CEO, David Villalón, says they're doing things differently. Instead of just using AI to generate responses, they're using it to build the whole process needed to get to the response. They call it "chain-of-work." Think of it like this: instead of just giving you the answer, the AI figures out all the steps needed to get there.

The brains behind this "chain-of-work" is Manuel Romero, Maisa's co-founder. He and Villalón previously worked together at another AI startup. They saw firsthand how unreliable AI could be and decided to build a solution. So, in 2024, they teamed up to tackle those pesky AI "hallucinations," where the AI makes stuff up. You know, when it confidently tells you something that's completely false.

How Maisa AI Works

They're not saying AI is bad, but they know that it's impossible for humans to check every single thing AI does. To fix this, Maisa uses a system called HALP, which stands for Human-Augmented LLM Processing. It's like having a bunch of students at a whiteboard. The digital workers ask you what you need, and then they outline every step they'll take to get there. It ensures that they are on the right track and that the outcomes will be satisfactory to all parties involved.

They also created the Knowledge Processing Unit (KPU). It is a system that helps to reduce those AI hallucinations. Maisa realized that being trustworthy and accountable was a big deal for companies that wanted to use AI for important tasks. They already have clients like a big bank, car manufacturers, and energy companies using their system.

Basically, Maisa wants to be a more advanced version of robotic process automation (RPA). They want to help companies boost productivity without needing a ton of rigid rules or manual programming. To help with this, companies can either use Maisa's secure cloud or deploy the system on their own servers.

Because they're focused on enterprise clients, Maisa doesn't have millions of users yet. But they're hoping to grow their customer base with Maisa Studio. They also want to expand with existing clients who have operations in different countries. With offices in Valencia and San Francisco, Maisa already has a presence in the U.S.

It looks like Maisa is on the right track, focusing on complex situations where companies want AI to be reliable and accountable. They have competitors, of course, but their focus on trustworthiness could give them an edge.

Maisa plans to use the new funding to grow its team and meet the growing demand. They expect to see a lot of growth soon as they start working with companies on their waiting list. David Villalón says they're going to show everyone that they can deliver on the promises of AI.