Enterprise automation has reached a decisive juncture. A study from MIT’s NANDA initiative recently reported that 95 percent of automation pilots stall before reaching meaningful production use. The figure shows a striking pattern: projects launch with great enthusiasm but lose traction when organizations attempt to implement them into daily operations. The recent announcement that Maisa, a Valencia- and San Francisco–based startup, secured $25 million in seed funding stands out as a test case for how companies are attempting to alter this pattern.
Rethinking Automation Workflows
Founded in 2024, Maisa has built its model on the idea that enterprises require accountable digital workers, not black-box systems. Its flagship product, Maisa Studio, enables organizations to design what the company calls a “chain of work.” Instead of focusing on single responses, these digital workers outline each step of a process before execution, creating visibility for review and correction.
Chief Executive David Villalón explained the rationale in an interview: “Instead of using technology to build the responses, we use it to build the process that needs to be executed to get to the response”. By foregrounding the process rather than the outcome, Maisa positions itself closer to traditional workflow management, but with greater flexibility.
Villalón and co-founder Manuel Romero, the company’s Chief Scientific Officer, previously worked at Spanish startup Clibrain. There they observed persistent errors in text-generation projects and concluded that enterprises needed tools to verify steps before approving final results and that is where this idea emerge. Their experience shaped the decision to design a model where humans remain central, guiding and checking digital workers throughout.
The Mechanics of Accountability
Maisa’s approach rests on two main technical components. The first is HALP, or human-augmented processing, which ensures users are prompted to review and approve stages of work. Villalón compared it to “students at the blackboard,” where every step is written out before the solution is accepted.
The second is the Knowledge Processing Unit, a deterministic system that reduces unwanted deviations. Together, these elements serve a single purpose: making the work explainable. While many companies still chase speed and novelty, Maisa argues that enterprises cannot adopt automation at scale without clear accountability.
Clients in banking, automotive manufacturing, and energy have already begun using the system in production. These sectors often require audit trails and compliance checks, making them fertile ground for stepwise methods. Maisa also offers both cloud-based and on-premise deployments, catering to industries that must maintain strict control over sensitive data.
Positioning in a Competitive Market
Maisa’s model sits within a crowded field of automation startups, many of which also promote digital workers or workflow agents. Competitors such as CrewAI have emphasized rapid execution and low-barrier experimentation. Others focus on visual coding or simplified scripting for business teams. Where Maisa diverges is in prioritizing oversight from the outset. Its product is designed not only to execute tasks but also to document how they are executed.
This orientation aligns it less with consumer-facing tools and more with enterprise governance systems. In that respect, Maisa resembles a next phase of robotic process automation (RPA), with the difference that its digital workers learn from instruction rather than rigid scripts. Whether enterprises will prefer verifiable process chains over the faster, more flexible outputs of rivals remains an open question.
Seed Round and Workforce Growth
The $25 million seed round was led by Creandum, a European venture firm, with participation from Forgepoint Capital International and Spanish bank Banco Santander through their joint venture. The round follows a $5 million pre-seed led by NFX and Village Global in 2024. Maisa intends to grow its workforce from 35 to around 65 by early 2026 with this financing.
The company has also announced plans to expand with existing customers that operate across multiple countries. Maintaining dual headquarters in Valencia and San Francisco gives Maisa a presence in both European and U.S. markets, a practical move as regulatory and enterprise requirements differ across jurisdictions.
Addressing Persistent Automation Failures
Maisa’s rise coincides with an industry-wide shift. For years, automation has been tested in narrow pilots that rarely advanced. These failures eroded confidence but also clarified where projects went wrong: lack of transparency, absence of review, and weak governance. Enterprises now appear to be demanding traceable processes rather than outputs that cannot be explained.
In that light, Maisa’s emphasis on accountability signals a larger correction. The startup’s model resonates with organizations accustomed to audits, compliance reviews, and stepwise approvals in finance or supply chain management. If digital workers are to play a lasting role in these industries, their processes must be as visible and controllable as existing enterprise systems.
Reliability as a Strategic Direction
In a LinkedIn post, Villalón remarked that the “quick start often becomes a long nightmare when reliability and the ability to correct mistakes are missing.”
His observation points to a challenge widely recognized across the enterprise landscape. Systems that deliver speed at the outset often reveal structural weaknesses when errors emerge without a clear path to correction. Maisa’s position is that process chains, validated step by step, offer a means of maintaining transparency and reducing the risk of systemic failure.
The company’s leadership has also framed its expansion in straightforward terms. “We are going to show the market that there is a company delivering what has been promised, and that it is working,” Villalón said. This stance is less about promoting technology for its own sake and more about matching enterprise expectations with accountable systems.
