Partnership: Lufthansa Industry Solutions and Novagentica bring AI agents safely into enterprise systems

Norderstedt/Zurich, November 6, 2026 – Lufthansa Industry Solutions (LHIND) and Swiss software company Novagentica are entering into a solutions and implementation partnership. Together, they aim to move their customers’ AI agents from pilot projects into production, where they actively work in ERP, CRM, service and logistics systems without compromising control or compliance.

Novagentica provides Architect™ as a licensed software platform. LHIND integrates it into its customers’ existing system landscapes and processes and subsequently operates it as a managed service. The IT consulting company draws both on its process expertise from the highly regulated and safety-critical aviation sector and on its industry expertise in logistics and mobility, automotive, retail and manufacturing to accelerate and streamline the path to production.

Both partners see the same barrier among their customers: as long as AI answers questions or reads data, the path into operations is short. The moment it is expected to act as an agent, independently booking, ordering or changing data in systems of record, clearly defined decision rights and reliable evidence of what the agent actually did become essential.

Traceability instead of a black box

Novagentica Architect™ enables enterprises to centrally engineer, deploy, operate and govern AI agents, independently of runtime and model. Execution remains on the customer’s chosen platform, such as Microsoft Copilot Studio, Salesforce Agentforce, AWS Bedrock, Google Vertex, LangGraph or an in-house framework. Enterprises stay flexible and avoid lock-in to a single provider.

This matters most for business-critical actions such as a purchase order, a credit note, a price change or a container rebooking. Technically, the approach follows a well-established principle: whoever executes does not verify. Agents hold no privileged system credentials of their own. Consequential actions are authorized in advance against a certified “Decision Contract,” a machine-enforceable set of rules covering value limits, approval thresholds, escalation paths and prohibited decisions.

After execution, the result is independently verified against the data in the system of record. A runtime’s success message is therefore no longer the only evidence available. The resulting chain of evidence is designed, among other things, to support the documentation requirements of the EU AI Act.

Four real-world examples

• Automotive

  Agents support product development for a vehicle project, with the goal of cutting development time from 48 months to 12 to 18 months.

• Transport and logistics

  Agents monitor shipments and independently rebook containers when disruptions occur.

• Industrial manufacturing

  Agents continuously assess suppliers and initiate a switch before a disruption reaches production.

• Consumer goods

  Agents identify quality issues, trace the root cause back to the batch and supplier, and monitor corrective action through to completion.

From pilot project to production

“Our customers are no longer asking whether AI agents work, but how they can deploy them responsibly in production,” says Rainer Liebhart, Managing Director of LHIND. “Together with Novagentica, we help companies across multiple industries move successful AI pilots into production and embed Agentic AI into their value-creating activities for the long term.”

“Developing an AI agent is becoming increasingly easy. Getting it to act in a controlled manner is the real challenge,” adds Ralf Struckmeier, Vice President Markets at LHIND. “This is precisely where we support our customers, at the intersection of technology, processes and operational responsibility.”

“An agent that orders, books or assesses is not something you click together, it is something you engineer,” says Martin Hofmann, Founder of Novagentica. “A runtime can report what it sent, not what actually arrived in the system. That interface is precisely where our strength lies.”

“We keep hearing the same sentence: the agent works, but it isn’t allowed near the system,” says Felix Bühner, Chief Revenue Officer of Novagentica. “That is the blocker we are removing together with LHIND, as a repeatable approach that can be transferred from aviation to manufacturing.”

Three areas are initially in focus: deploying agents in production with controlled write access to systems of record, a joint compliance-by-design offering aligned with the EU AI Act, and applications in industry and logistics. In an initial joint customer project, the partners are already implementing controlled write access by an agent to a system of record and developing this into a standardized, scalable approach that can be applied across aviation, logistics, mobility, manufacturing and retail.

About Lufthansa Industry Solutions

Lufthansa Industry Solutions is an IT consulting and system integration service provider. The Lufthansa subsidiary supports its customers in the digital transformation of their businesses. Its customer base includes both companies within the Lufthansa Group and more than 300 companies across a range of industries. Headquartered in Norderstedt, the company employs more than 3,000 people at locations in Germany, Albania, Switzerland and the United States.

About Novagentica

Founded in 2025, Novagentica is a Swiss software company with a presence in Berlin. Its founder, Martin Hofmann, is the former Group CIO of Volkswagen AG. With Novagentica Architect™, the company provides a licensed platform that enables enterprises to engineer, operate and govern AI agents independently of runtime and model, with a focus on applications in which agents trigger business-critical actions in enterprise systems.

Press contacts

Lufthansa Industry Solutions

Heiko Packwitz
Corporate Communications

Novagentica

Maru Winnacker
Corporate Communications