Maisa AI, a startup focused on building an artificial intelligence agent platform, today announced that it has successfully raised $25 million in seed funding.
The investment was led by Creandum, an early backer of Spotify Technology SA. Forgepoint Capital International, NFX, and Village Global also participated in the round. The latter two funds previously invested $5 million in Maisa AI in December last year.
The platform enables business users to create AI agents through natural language prompts. According to the company, its software can be used without programming expertise or training datasets. Agents created via the platform can perform tasks such as extracting data from forms and scanning network traffic to detect malicious activity.
"Users can scale AI quickly and securely without relying on the support of an entire development team," said David Villalon, co-founder and CEO of Maisa AI.
The company stated that its AI agents are capable of generating step-by-step explanations detailing how they complete each task. This allows users to verify whether the work has been done correctly. If an agent makes a mistake, employees can provide natural language feedback indicating where improvements are needed.
Maisa AI’s platform can automatically overcome certain processing obstacles. The software includes a so-called self-contained mechanism that allows agents to detect changes in task requirements and adjust accordingly. Furthermore, if an agent cannot complete a task based solely on the user’s initial prompt, it can request additional instructions or access third-party applications.
In the background, Maisa AI's platform is powered by a software engine called KPU. When employees ask an agent to perform a task, KPU uses a large language model to plan how the task should be executed. It then sends the plan to a component that carries out the task and generates feedback on how to perform it more efficiently in the future.
Last March, Maisa AI first detailed KPU and launched an updated version called Vinci KPU in November. The upgraded version offers lower latency and supports compute-on-demand during inference, meaning it can dynamically scale infrastructure during prompt processing to improve output quality.
Vinci KPU also includes what Maisa AI describes as an "integrated knowledge base consultation system." This mechanism allows the software to verify the accuracy of AI agent responses. The company claims that Vinci KPU can locate task-critical information more cost-effectively than RAG (retrieval-augmented generation) tools.
Maisa AI says its platform has already been adopted by several large enterprises, including banks, automotive manufacturers, and energy companies. The company plans to use the newly raised funds to expand its international presence and hire additional talent.
