Interloom raises $16.5M to give AI agents enterprise memory and automate real-world workflows

AI agents are starting to move into core business operations. There’s just one problem: they don’t remember how work actually gets done.

That gap is what Interloom is going after. The Munich, Berlin, and London-based AI startup has raised $16.5 million in a seed round led by DN Capital, with participation from Bek Ventures and Air Street Capital. Its pitch is simple but ambitious—turn employees’ day-to-day decisions into a living memory layer that AI agents can rely on.

Most companies run on knowledge that never makes it into documentation. It sits in inbox threads, support tickets, internal tools, and the heads of experienced employees. When those employees leave, a large part of that knowledge goes with them. AI systems, for their part, can follow instructions but often lack the context needed to make consistent decisions in messy, real-world scenarios.

Interloom’s platform watches how teams handle real operational cases across systems and captures those decisions as structured memory. Instead of asking AI agents to rely on static playbooks, the system feeds them patterns drawn from actual outcomes. Over time, that memory grows and reflects how the organization truly operates.

With $16.5M in funding, AI startup Interloom aims to power AI agents with enterprise memory for automation

The company frames it as a shift from documentation to experience. Rather than writing down every possible rule, teams build a record of what has worked in the past. AI agents can then act on proven resolutions, and employees can reuse past decisions without starting from scratch.

Interloom says the platform is already in use at large enterprises, including Zurich Insurance and Volkswagen. In those environments, it processes millions of operational cases, helping teams resolve issues faster and automate workflows that previously required constant human judgment.

“An agent is only as good as the specific knowledge it can rely on. The problem is that context is dynamic, poorly documented and lives in the daily decisions of expert front-line workers. Interloom stood out by building a corporate context graph that continuously captures real-world decisions and how organisations actually operate,” said Guy Ward Thomas, Partner at DN Capital.

Founder and CEO Fabian Jakobi sees this as a missing layer in enterprise AI. “AI agents are moving into core business operations, but without company-specific memory, they can’t act reliably,” he said. “We ground their decisions in real past resolutions, so automation is based on proven outcomes, not assumptions.”

The company plans to use the new funding to expand across more enterprise customers and push its go-to-market efforts. It is introducing what it calls a “Chief of Staff” agent, designed to help organizations roll out AI automation across complex workflows by coordinating tasks across systems and teams.

Interloom’s approach lands at a time when enterprises are testing how far AI agents can go beyond chat interfaces and into execution. The challenge is no longer just generating answers—it’s making decisions that hold up in production environments.

If Interloom can turn fragmented operational knowledge into something machines can reliably act on, it could give companies a way to preserve institutional memory and make automation less brittle. The bet is that the future of enterprise AI won’t be driven by better prompts, but by better memory.