Marble Cognition 获 375 万美元 pre-seed 融资，用 AI 智能体观察并学习员工工作流程

Artificial intelligence agents have become two a penny in the enterprise today, but just because they’re rapidly being commoditized does not mean organizations know how best to use them.

That’s the challenge a startup called Marble Cognition is trying to solve after raising $3.75 million in pre-seed funding to coincide with its launch today. The round was led by Heartcore and Seedcamp, and saw participation from angel investors that include executives from McKinsey & Co., Mistral AI and a board member of Datadog Inc.

Munich-based Marble was co-founded by the German AI engineers Felix Rösner (pictured, left) and Christophe Schmit (right), who were among the first to realize that AI agents have already become a commodity, used by almost every enterprise in some capacity or other. Most organizations are using AI agents primarily for writing software, but Rösner believes that to see the real value of those autonomous workers, companies need to provide them with the granular operational context that tells them what they actually need to build and automate.

But building this picture is a notoriously difficult challenge because few corporate workflows run according to neat blueprints. Though a study by McKinsey & Co. suggests that it’s theoretically possible to automate 57% of all work hours in the U.S., most companies haven’t gotten anywhere near that goal.

That’s because more than 90% of business processes live within things like messy email threads, scattered spreadsheets and inside the minds of employees. To teach AI agents what needs to be built and automated, companies need to piece all of those processes into a picture they can follow.

This is exactly the challenge Marble seeks to address. Its AI agents are designed to capture those business workflows at the source. The startup has developed a dedicated desktop application that sits and monitors how employees work using different tools and processes. It takes screenshots and captures worker’s clicks and keystrokes, before transforming this data into a structured process map that AI agents can read and execute, similar to code.

“Every company we talk to wants AI to take over real work,” Rösner said. “But nobody can tell the AI how that work is actually done. So we solve that first. Our Coworkers watch how the work happens before they take it over. They do it the same way your best people do.”

In case employees or their bosses are worried about privacy, Marble reassures them that all of the data it records of their workers working day-to-day is anonymized and can be reviewed before it touches any AI models. In addition, all of its model’s processing can be done within the customer’s private cloud or on-premises environment. It learns to identify non-business-associated computer use too, so if an employee is browsing Facebook during a lunch hour, it will exclude that from the data it collects.

Once the data is gathered, Marble then deploys its Coworkers, which are designed to take over the most repetitive work processes from people and run them end-to-end an organization’s existing software stacks, with humans remaining in the loop when important decisions need to be made.

Rösner said Marble has already delivered some serious efficiency gains for early adopters like EGYM GmbH, Mutares SE & Co. KGaA and Enam GmbH. In the case of one private equity-backed U.S. financial services firm, Marble’s Coworkers were configured to create settlement offers for human review. As a result of that work, the company was able to slash its case-handling times by 90%, which meant that its human staffers could process around 50% more claims without needing to hire anyone else.

Seedcamp General Partner Reshma Sohoni, an early backer of the robotic process automation firm UiPath Inc., said the vast majority of companies in the world want to learn how to automate more of their work. “The hard part has always been mapping out how the work gets done, so you can actually automate it,” he said. “Marble fills that gap, giving companies vital context for AI transformations that could define their future.”

Photos: Marble Cognition