Marble Cognition Raises $3.75M for AI Agent Workers That Learn by Watching Employees

Marble Cognition Secures Pre-Seed Funding to Transform Enterprise Automation

Munich-based startup Marble Cognition has raised $3.75 million in pre-seed funding to address a critical gap in enterprise AI adoption: teaching AI agent workers how work actually gets done1

1

. The round was co-led by Heartcore Capital and Seedcamp, with participation from angel investors including a board member of Datadog, executives from McKinsey, Mistral AI, and Lovable, plus a founder of Personio2

2

. Founded in 2025 by German AI engineers Felix Rösner and Christophe Schmit, the company plans to use the capital to scale inference capabilities, expand its engineering team in Munich, and launch operations in the United States2

2

.

AI Agent Platform Captures Hidden Business Processes

While AI agents have become commonplace in enterprises, Marble Cognition tackles the harder problem of capturing the operational context that tells these autonomous workers what to build and automate. A McKinsey study suggests 57% of all work hours in the U.S. could theoretically be automated, yet most companies fall far short of this goal because more than 90% of business processes exist in messy email threads, scattered spreadsheets, and inside employees' minds1

1

. The agentic AI start-up developed a dedicated desktop application that monitors how employees work using different tools and processes, taking screenshots and capturing clicks and keystrokes before transforming this data into structured process maps that AI can read and execute1

1

.

How Marble's AI-Driven Automation Works

The platform observes teams at work and converts every workflow into a system AI can interpret like code, then deploys what Marble calls 'Coworkers' to automate manual white-collar tasks end-to-end within existing software stacks2

2

. These AI agent workers handle repetitive tasks while keeping humans in the loop when judgment matters. "Every company we talk to wants AI to take over real work. But nobody can tell the AI how that work is actually done," said Felix Rösner. "Our Coworkers watch how the work happens before they take it over. They do it the way your best people do"1

1

. The desktop application records activity only in business apps during working hours, never in private apps or tabs, and employees can pause recording at any time2

2

.

Privacy Safeguards and On-Premises Deployment

All data captured by Marble's platform is anonymized before any person or AI model accesses it, and the system analyzes information in aggregate rather than tracking individuals2

2

. Employees can review collected data before it reaches any AI models, and all processing can occur within a customer's private cloud or on-premises environment1

1

. The platform also learns to identify non-business computer use, excluding activities like browsing Facebook during lunch breaks from its data collection1

1

.

Early Adopters See Dramatic Efficiency Gains

Marble Cognition already works with tech unicorns, private equity-backed companies, and AI roll-ups including EGYM, Mutares, and Enam2

2

. At one private equity-backed U.S. financial services firm, Marble's Coworkers were configured to create settlement offers for human review, slashing case-handling times by 90% and enabling staff to process approximately 50% more claims without additional hiring1

1

. These efficiency gains demonstrate how the platform helps organizations automate enterprise workflows that were previously too complex to map and systematize.

Investor Perspective on Workflow Automation Gap

Reshma Sohoni, general partner at Seedcamp and early backer of robotic process automation firm UiPath, emphasized the challenge Marble addresses: "Most companies want to automate more, but the hard part has always been mapping out how the work gets done. Marble fills that gap, giving companies vital context for AI transformations that could define their future"2

2

. Max Niederhofer, partner at Heartcore, added that the Munich-based startup is building "a key technology for the AI-enabled workplace" by learning entire company processes, anonymizing data, deriving standard operating procedures, and deploying agentic AI against them2

2

.