INXM Exits Stealth With €5.7M to Make Enterprise AI Reliable and Auditable

Who Founded INXM

The company was founded in 2025 by Alex Oelling, Matthias Kainer, Jesper Bylund, and Kamil Klüber . The founding team brings direct experience from industries where process reliability is non-negotiable. Oelling, Kainer, and Bylund served as Chief Digital Officers or data leaders at Isar Aerospace and Volocopter, while Klüber is a veteran of n8n and Thoughtworks . That background — aerospace manufacturing, urban air mobility, open-source workflow automation, and enterprise software consulting — shaped both the technical architecture and the target market.

What Makes Compiled AI Different

The core problem with standard LLM-based agents is that they reinterpret instructions at transaction time. The same input can produce different outputs on different runs, and there is no built-in audit trail for regulatory scrutiny. Models can also hallucinate steps or introduce unpredictable edge cases, which is unacceptable when a workflow crosses ERP, PLM, MES, and factory-floor systems .

Compiled AI inverts that model. Instead of calling an LLM every time a workflow runs, INXM uses the LLM during a one-time “compilation” phase. Natural-language requests are converted into deterministic, executable code. That code is validated through multi-stage checks and then deployed. Once compiled, the workflow executes without any further LLM inference — producing the same result every time it runs, with full logs and explicit human sign-off points for critical actions .

The process mirrors the difference between compiled software and interpreted prompts. By separating the intelligent generation step from the repetitive execution step, INXM achieves the flexibility of natural-language instruction with the repeatability of traditional deterministic code. Because the Orchestrator runs on the enterprise’s own infrastructure rather than on external cloud servers, sensitive operational data never leaves the building — a hard requirement for GDPR and the EU AI Act compliance .

Target Industries and the German Mittelstand

INXM is targeting enterprise operations, with a specific emphasis on the German Mittelstand — the country’s backbone of mid-sized industrial firms. These companies often operate complex software stacks that span ERP, PLM, MES, and shop-floor systems running on private servers that cloud-based automation tools cannot reach .

Manufacturing is the starting point. In these environments, a single production-order change can require coordinated updates across half a dozen systems, with managers needing to approve each step. Compiled AI lets organizations design a process once in natural language, compile it, lock it into a reproducible workflow, and then run it reliably — with human-in-the-loop approvals where needed .

What the €5.7 Million Funds

INXM’s immediate plan is to deploy Orchestrator in its first enterprise accounts and scale the platform. The company is taking a top-down approach, targeting decision-makers who don’t need to involve an engineering team to automate cross-system processes . The founding team’s experience at aerospace and mobility companies — where a single workflow failure can cascade into production delays — informs both the product’s architectural bets and the go-to-market focus on manufacturers and other regulated industries that require repeatable, auditable automation .