AI startup Marble Cognition has raised $3.75 mn in a pre-seed funding round co-led by Heartcore and Seedcamp to develop technology that identifies and automates undocumented business processes.
The round also attracted angel investors, including a Datadog board member, a Personio founder and executives from McKinsey, Mistral AI and Lovable.
- Marble Cognition secured $3.75 mn in a pre-seed round co-led by Heartcore and Seedcamp, with backing from investors associated with Datadog, Personio, McKinsey, Mistral AI and Lovable.
- The startup develops AI agents called Marble Coworkers that observe employee workflows, map business processes and automate tasks using existing enterprise software.
- The funding will support improvements to Marble’s inference capabilities, expansion of its Munich engineering team and entry into the US market.
Founded in late 2025 by Felix Rösner and Christophe Schmit, Marble is developing AI-powered software that helps organisations understand how their internal operations function before introducing automation.
The company is addressing a common obstacle to enterprise AI adoption: many business processes are not formally documented, with operational knowledge scattered across emails, spreadsheets and individual employees.
This lack of documentation makes it difficult to determine which tasks are suitable for automation or provide AI systems with the instructions needed to execute them.
Marble develops AI agents to automate undocumented workflows
McKinsey estimates that existing technology could theoretically automate approximately 57% of working hours in the US. Yet businesses often lack the structured process information required to implement such automation.
90% of core business processes at the companies it works with remain undocumented.
Its platform analyses how employees perform everyday tasks and converts their activities into structured process maps that AI systems can interpret.
Using these maps, Marble creates AI agents known as Marble Coworkers, which execute business processes through software tools already used by the organisation. Employees remain involved where tasks require human judgement or approval.
“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.
Desktop application maps employee workflows with privacy controls
Marble collects workflow information through a desktop application that observes activity across business software, including screen interactions, mouse clicks and keystrokes.
The collected information is used to reconstruct operational processes rather than assess individual employee performance.
According to Marble, data is anonymised before employees or AI models can access it, and analysis is conducted at an aggregate level rather than for individual workers.
The application excludes private applications, private browser tabs and activity outside working hours. Employees can also pause monitoring whenever necessary.
Marble said its technology has already been deployed at several organisations, including EGYM, Mutares and Enam, alongside other technology businesses and private equity-backed companies.
The startup plans to use the new investment to scale its inference capabilities, recruit additional engineers in Munich and support its launch in the US market.
