KI-Agenten statt Forward Deployed Engineers: 3,75 Mio. Pre-Seed – Felix Rösner (Marble Cognition)

Marble Cognition bets AI co-workers replace enterprise consultants

Enterprise AI has a deployment problem. Companies cannot describe their own work, so they hire Forward Deployed Engineers to sit beside employees and map processes that go stale once those engineers leave. Marble Cognition's bet is to capture that understanding from customer data and put AI co-workers directly inside the company instead of selling tools to the engineers who implement software.

Takeaways

• Marble deploys co-workers instead of implementation tools. Marble Cognition captures how work actually happens from customer data and puts its agents inside the company, rather than selling software to the engineers who implement it. It went from two or three customers to about a dozen.
• Process documentation is stale or missing. Rösner says the first question to a customer is whether their processes are documented. The answer is usually no, or the documentation is years out of date. In one process up to 20 people had each built their own method over decades.
• Forward Deployed Engineers fill the gap. Standard AI products were not customisable enough for large, older companies, so hundreds of Forward Deployed Engineer roles appeared on LinkedIn. Rösner compares the work to McKinsey-style consulting, and says the maps they build go stale once they leave.
• Marble raised quietly and launched on its terms. The 3.75m dollar pre-seed closed months before Marble announced it, and the round itself took about a week and a half. Rösner argues a launch is a special event, so the team spent the summer on hiring and product instead.
• The model depends on land and expand. Marble lands with a single use case and shows fast time to value, then expands across a horizontal set of use cases. Rösner says expansion is central to the model and creates stickiness.
• Munich keeps top-talent costs below the US. Marble stayed in Munich for its deep talent pool, which the founders already knew through the CDTM network. Rösner says competition for top people and salary benchmarks are lower there than in San Francisco or New York.
• Rösner picked angels for four distinct jobs. Founders serve as mentors he can call for guidance, distribution angels come from consulting and private equity, super angels are wired into venture capital, and technical angels include people at Mistral and a Datadog board member. He wants to be WhatsApp-close with each.

What comes next for Marble

• Push go-to-market and hit Q1 milestones on net revenue retention and net new ARR.
• Expand into the US and build a best-in-class product rather than one for a regional market.