With version 7.0, Zammad responds to the growing demand for AI assistance in customer support. As requirements for data protection and compliance—as well as preferences for specific AI models—vary significantly across industries, Zammad consistently focuses on openness and freedom of choice when it comes to language models.
At the core of this strategy is the new AI API, which allows organizations to choose which language models to use. Supported options include well-known cloud services, such as OpenAI, Anthropic Claude, and Google Gemini, as well as European alternatives, such as Mistral AI. Organizations with strict requirements regarding data processing, storage location, and access control can run open-source models, such as Meta Llama, in a fully self-hosted environment.
"Companies should not be forced into an impossible choice between adopting AI technology and protecting their data," says Martin Edenhofer, founder and CEO of Zammad. "They should be able to decide where AI is used, which data is processed, and in which environment. That decision-making power is exactly what we put at the center. With Zammad 7.0, we provide the technical foundation for organizations to deploy AI in a controlled manner that aligns with their individual security and compliance requirements.”
A central aspect of Zammad’s AI approach is its alignment with the EU AI Act. With its transparent architecture and freedom to choose an underlying language model, Zammad allows organizations to proactively meet strict regulatory requirements regarding risk management and governance. Since Zammad prioritizes AI assistance over full AI autonomy, the principle of human oversight ("human in the loop") remains intact.
The auditability of the open-source codebase ensures that the technology aligns with European standards for trustworthy AI. Consequently, organizations have the necessary tools to fulfill their transparency obligations while maintaining complete control over their data processing environments.
Zammad 7.0 focuses specifically on AI support that helps support teams in their day-to-day work. The AI functions are built to reduce manual effort while leaving decision-making in human hands. Key features include:
- AI ticket summary:
Complex and lengthy communication threads are condensed into clear, structured summaries, making it easier for teams to take over tickets and reducing onboarding time when agents step in.
- AI Writing Assistance: Support agents with drafting and refining replies, including tone suggestions and spelling and grammar corrections. Final approval always remains with the support agent.
- AI agents: Automated categorization, prioritization, and routing of requests help teams quickly classify and process incoming tickets more efficiently.
The goal is to shorten handling times, simplify handovers, and improve the overall quality of customer communication. At the same time, AI only accesses ticket content within the defined context of a task. There is no uncontrolled sharing of data.
Despite the integration of AI, Zammad remains fully committed to its open-source promise: the source code is completely open, and the architecture is transparent and auditable. Zammad can be deployed as usual on-premises, in an organization’s own data center, or within controlled cloud environments, even when AI features are enabled.
"With Zammad 7.0, we aim to show that AI and digital sovereignty can coexist," says Gerrit Daute, Product Owner at Zammad. "Our goal is to provide a platform that adapts flexibly to our customers' individual security requirements, whether they rely on global cloud models or private instances."
Zammad is one of the first open source help desk solutions for customer and IT support that offers AI functions with a free choice of large language model (LLM). The platform bundles requests from email, phone, chat, web, and other channels in a central interface and can be flexibly integrated into existing IT and support landscapes. With a strong focus on transparency, control, and digital sovereignty, Zammad enables organizations to run modern support operations on their own terms.