Agent-based AI
for robust enterprise software
AI in Production: Selected Projects
Select a use case tab to explore.
ESACA (European Sovereign Agentic Coding Appliance)
Problem:
The public sector and regulated industries want agentic coding, but not if it means sacrificing data sovereignty. Code, prompts, and telemetry must remain within European infrastructure and not end up in U.S. clouds.
What's needed is an auditable, European-sovereign, turnkey, on-premises stack that improves continuously with each deployment rather than becoming stagnant.
Solution:
ESACA delivers a turnkey AI appliance designed for European sovereign deployments. It includes Fujitsu PRIMERGY server hardware with a GPU, an on-premises LLM, an agent layer, audit logging, MCP integration, and air-gap capability, all on the A12 platform.
The appliance is agentically developed itself: an Obsidian-based knowledge base (a structured, locally hosted documentation environment) autonomously expands after each use, creating a self-improving feedback loop. This means that ESACA gets smarter with every deployment without the need for manual retraining.
ELSTER, Germany's national tax filing platform and one of Europe's most demanding e-government environments, serves as the primary evaluation environment. Seven Agentic Engineering, mgm's methodological framework for agentic software development, provides the structural backbone.
Cosmo — AI backbone for insurance companies
Problem:
Insurers don't need AI as an isolated chatbot; rather, they need it as a backend for email, underwriting, damage assessment, document management, and inventory processes. It should be integratable into existing insurance platforms and the Microsoft world of material processing.
Solution:
Cosmo is mgm's AI backend for insurance. It is model-agnostic and can be integrated with other systems. It is built for email assistance, document processing, and accelerating processes across the full insurance lifecycle.
As the first inventory management system natively connected to Microsoft Copilot, Cosmo brings insurance data and logic directly into the M365 interface that teams already use — eliminating the need for context switching or separate tools.
Cosmo is actively used by leading insurance organizations such as SCOR, Roland, Allianz, and HDI. Access to the reference environment is available upon request.
Förderfinder: AI-Assisted Application Summary
Problem:
Funding programs generate large volumes of applications. Manually reviewing, understanding, and comparing each one is time-consuming and slows down processing. Specialists waste time digesting each application before beginning the technical assessment.
Solution:
The AI-assisted Förderfinder automatically converts descriptions of funding programs from unstructured data sources into the standardized XFLB format. This makes essential information immediately accessible in a consistent structure. Tasks that used to require manual extraction and reformatting now happen in seconds.
Its architecture combines deterministic logic for extracting structural data with LLMs for language and context, and it is backed by an evaluation framework that quality-checks every summary before it reaches the reviewer.
The Förderfinder was developed as an once-only solution under the direction of the Free State of Bavaria and is available for reuse by all German federal states and the federal government. This lays the groundwork for a nationally standardized, fully searchable funding landscape.
AI-first SaaS ERP Platform for Institutional Catering Management
Problem:
A leading provider of catering management software needed to completely reimagine its product. The new platform had to be a modern, API- and AI-first SaaS ERP platform with out-of-the-box data structures and 24-hour onboarding. The new platform had to be production-ready in three month — not as a prototype, but as a marketable product.
Solution:
mgm developed the SaaS ERP platform from scratch using its Agentic Software Engineering approach throughout the entire software development lifecycle. A lean team of fewer than eight full-time equivalents (FTEs) organized as business-tech tandems replaced an AI-assisted codebase representing more than four developer years within ten days. The team then delivered a productive enterprise platform with 250,000 lines of code within four months.
The result: an enterprise ERP platform ready for market launch in sixth months, including the first pilot customers.
The architecture combines high automation through multiple AI services with deterministic calculation logic, delivering speed without sacrificing reliability — 2-5 times faster than AI-assisted coding.
ESACA:
Agentic Coding — On-premises, Under your Control.
Tailor-made Enterprise AI – From the initial idea to scalable AI integration
We help companies implement AI in a targeted manner, from the initial idea to scalable AI integration.
Modular, process-oriented, and measurable.
READINESS
TRANSFER
This clarity gives rise to a target vision that provides guidance for the strategic use of AI in your company. On this basis, use cases can be prioritized according to their business benefits and translated into an actionable roadmap. This gives managers the transparency and clarity they need to take responsibility and set the course for the next steps.
This is particularly valuable when it is still unclear where AI can create the greatest added value in your company or what conditions need to be created within the company.
- Clarity of strategic objectives
- Discovering areas for impact
- Defining Benefits for Business Processess
We support you in creating the right setting for roles and responsibilities in the form of an interdisciplinary committee that takes over strategic orchestration and anchors the framework in governance and processes. At the same time, we prepare the organization for change by establishing initial approaches for upskilling and change management at an early stage.
This step is particularly useful when initial use cases are becoming clearer, but it is not yet clear how they can be scaled securely and compliantly. This framework is a resilient foundation that goes far beyond individual prototypes and enables the scaling of future use cases.
- Establishing Governance
- Setting Strategic Direction
- Checking Feasibility
This allows you to quickly identify which approaches are viable and should be prioritized for the next implementation phase. At the same time, you lay the foundation for scalable AI implementation in your system landscape and your company, manage investments in a targeted manner, and minimize risks.
- Rapidly prototype and test solutions
- Gather actionable feedback
- Minimize risk
Using the mgm AI platform or your own systems, we identify technical, organizational, and regulatory potential, thereby continuously improving efficiency, quality, and transparency—ensuring that your AI delivers lasting, measurable business benefits and remains stable in operation.
- Measure achieved impact
- Validate results
- Optimize solution in a targeted manner
We help you build knowledge databases, regulate data access for authorized users, and create a secure platform. This enables us to detect attacks on the system at an early stage and respond immediately. Ideally, our collaboration begins with the creation of a target architecture before implementation begins. Even if technical questions are still open, we can quickly assess how sensitive your data is.
The end result is a sustainable system that meets legal requirements for data protection and security, protects your data efficiently, and avoids unnecessary steps.
- Seamless integration of data protection and compliance requirements
- Tailored to your security and protection needs
- Master knowledge building in the field of secure AI
This continuously develops the framework for the use of AI in your company and embeds it firmly in governance and processes – the basis for AI to not only have a selective effect, but to deliver measurable business benefits as an integral part of your organization and to truly achieve sustainable scaling.
- Transferring Learning Experiences
- Expanding Use Cases
- Strategically Anchoring AI
Companies and organizations that build with us:
mgm enterprise ai radar
Efficiency meets structure. With agents who understand what matters.
Less coordination. Faster implementation. Maximum efficiency.
Automated workflows instead of manual processing. Structured data. Immediately usable results.
- For services and data
- With flexible integration of AI functions
- Tailored to your security and protection needs
- Data sovereign, on-premises LLM
- Developed for the highest data protection requirements
- Deploys in your own infrastructure — cloud, data center, or hybrid
- Fully auditable and compliant by design
- Model-free and sovereign
- Sovereign Cloud from Germany
- BSI-C5 and ISO certified
- For secure hosting and full control in the sovereign cloud from Germany
- Infrastructure as Code
- Deploy in hours: with Terraform, GitOps, and automated environment configuration.
Check operating options now.
mgm AI insights
