Marble secures $3.75M to develop AI agents that learn from workplace activities
AI startup Marble Cognition (Marble) has secured $3.75 million in a pre-seed funding round jointly led by Heartcore and Seedcamp. The round also attracted support from several angel investors, including a Datadog board member, the founder of Personio, and senior executives from McKinsey, Mistral AI and Lovable.
Established in late 2025 by Felix Rösner and Christophe Schmit, Marble is building AI-powered technology that helps businesses discover, map and automate their internal workflows and operational processes.
With artificial intelligence becoming increasingly capable, businesses are exploring new ways to automate routine operations and improve productivity. However, many organisations still lack proper documentation of their internal processes. Important operational knowledge is often scattered across emails, spreadsheets, software platforms and employees' individual experience. As a result, identifying suitable tasks for automation and implementing AI systems effectively can be challenging.
Research from McKinsey suggests that existing technologies could potentially automate around 57 per cent of working hours in the United States. Marble also estimates that approximately 90 per cent of core business processes within the companies it works with remain undocumented, highlighting a significant obstacle to wider AI adoption.
Marble is developing a solution to bridge this gap by observing how employees complete their everyday responsibilities and transforming those activities into structured, easy-to-understand process maps that AI systems can use.
Using these insights, the platform creates AI agents called Marble Coworkers, which are designed to carry out business processes through the software and tools companies already use. The system also keeps employees involved in workflows that require human judgement, decision-making or approval, combining automated execution with human oversight.
Every company we talk to wants AI to take over real work. 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 way your best people do,
said Felix Rösner, co-founder of Marble.
The platform operates through a desktop application that captures how employees interact with different business tools, tracking activities such as on-screen actions, mouse clicks and keystrokes to understand existing workflows. Marble says that collected information is anonymised before it becomes accessible to employees or AI models, with analysis focused on aggregated patterns rather than individual employee behaviour. The software also excludes private applications, personal browser tabs and activity outside regular working hours. Employees can pause the monitoring process whenever they choose.
According to the company, its technology is already being used by several organisations, including EGYM, Mutares and Enam, alongside other technology firms and businesses backed by private equity investors.
The newly raised capital will help Marble strengthen and scale its AI inference infrastructure, grow its engineering team in Munich and accelerate its expansion into the US market.