Exclusive: Interloom, a startup leveraging tacit knowledge to empower AI agents, raises $16.5 million in venture funding

(SeaPRwire) –   Michael Polyani, a British-Hungarian philosopher, economist, and scientist, is widely recognized today for his concept of “tacit knowledge.” He observed that a significant portion of expertise in any field remains unwritten, existing as a form of professional intuition that even experts struggle to articulate. Polyani famously stated, “We know more than we can tell.”

This tacit knowledge poses a challenge for companies seeking to automate workflows using AI agents, as much of the necessary information is not formally documented.

Interloom, a Munich-based startup focused on modernizing business process automation for the AI era, believes it has a solution to the tacit knowledge problem. The company has successfully secured $16.5 million in venture capital funding to advance its mission.

The funding round was led by DN Capital, with contributions from Bek Ventures and existing investor Air Street Capital. Interloom had previously announced a $3 million seed round in March 2024.

Interloom has not disclosed its valuation following this latest funding round.

Fabian Jakobi, Interloom’s founder and CEO, contends that the current enthusiasm for AI agents overlooks the critical bottleneck of tacit knowledge. He estimates that approximately 70% of operational decisions are never formally recorded. For instance, when a complex support ticket reaches an experienced employee, they know how to resolve it, whom to escalate it to, and the correct solution not from a manual, but from prior experience.

“The most crucial individual in a bank is the one who can verify the accuracy of documentation,” Jakobi stated. “They are often the lowest paid, yet they are responsible for quality.”

Interloom’s methodology involves processing vast amounts of operational data, including support emails, service tickets, call transcripts, and work orders, to construct what it terms a “context graph.” This is a dynamic map illustrating how problems are actually resolved within an organization. Jakobi compares this to Google Maps: just as Google learns optimal routes from real-time traffic, Interloom maps the problem-solving paths taken by operational experts to guide both AI agents and new employees.

Jakobi is a seasoned entrepreneur, having previously founded Boxplot, which was acquired in 2021 by Hyperscience, an AI software firm specializing in extracting data from unstructured documents.

Interloom’s software is currently in use by several major European corporations. At Commerzbank, Interloom analyzed millions of customer support emails against existing internal documentation, revealing significant discrepancies and inconsistencies. The company reported reducing the gap between documented and actual operational knowledge from approximately 50% to 5%. Volkswagen is utilizing Interloom to process customer support tickets, and at Zurich Insurance, Interloom won a company-wide AI competition, outperforming an estimated 2,000 other AI startups, for an underwriting use case.

Jakobi explained that an underwriting decision at an insurance company reflects the firm’s specific risk tolerance, its historical experience with particular brokers and products, and institutional knowledge that generic AI models lack.

“The Zurich underwriter understands their broker chat underwriting process far better than a firm like Accenture,” Jakobi remarked, implicitly criticizing large consulting firms that have traditionally dominated enterprise process work.

The overarching argument is that AI agents, regardless of their sophistication, are ineffective in large enterprises without organization-specific context. Jakobi refers to this as the “corporate memory” challenge.

“In software development, a compiler confirms code functionality,” Jakobi noted. “We lack that luxury in other domains. Evaluation must come from a human expert.”

Interloom’s new investors share this perspective. Guy Ward Thomas, a partner at DN Capital, commented that “an agent is only as effective as the expert decisions it can draw upon.” Thomas added that DN Capital has observed in other AI agent startups that a lack of relevant enterprise context leads to poor performance. “Our experience with vertical AI agents and voice platforms has underscored the critical importance of context,” he said.

Mehmet Atici of Bek Ventures, who previously invested in UiPath, a leader in the prior wave of Robotic Process Automation (RPA), noted that RPA relied on agents programmed for rigid, repetitive workflows. “We have witnessed the transformative power of automation firsthand, and we believe AI is now ushering in a new era of rapid enterprise adoption,” Atici stated.

Interloom’s timing appears opportune, coinciding with the “Great Retirement” trend, where approximately 10,000 Baby Boomers retire daily in the U.S., taking with them decades of institutional knowledge precisely as companies are scaling their AI deployments.

Jakobi views the competitive landscape with characteristic directness, identifying inertia—the tendency within large enterprises to maintain existing operational practices—as his primary rival.

Interloom’s upcoming product, internally dubbed a “Chief of Staff,” is designed to provide managers with real-time insights into AI agent performance, including version control for agent-driven processes.

However, Interloom is not alone in developing AI agent management and orchestration solutions. Major AI providers such as OpenAI, ServiceNow, and Microsoft are also working on similar offerings.

Jakobi believes Interloom’s “context graph” provides a significant advantage over these larger competitors, who he argues often lack a comprehensive view across complex organizational processes.