German technology publisher Heise is among the first users of a new agent-enabled front-end layer that brings editorial logic directly into the system.
InterfaceAI is the front-end operational layer of InterRed, allowing editorial teams to automate their individual workflows directly within the CMS. Instead of using AI merely as a standalone tool, editorial logic is being integrated directly into the CMS.
The editorial team at Heise Online is already using the solution, including for a comprehensive fact-checking process.
Despite the growing use of AI, many editorial processes remain manual: structuring content, maintaining metadata, fact-checking or formatting texts, with AI frequently providing support only in specific areas, such as headlines or summaries, because without context, AI remains generic. It can write texts, suggest headlines and provide summaries, but it does not understand how an editorial team actually works.
As a result, AI frequently ends up as a standalone tool, rather than an integral part of editorial value creation. InterRed, a provider of an AI-based publishing automation solution for digital and print, aims to offer the answer to this structural problem with InterfaceAI.
A key difference from traditional AI integrations or API approaches such as MCP is that this layer integrates not just individual functions, but the entire editorial process logic. Publishers can map their individual rules, procedures and quality standards directly within the system, as an integral part of the daily workflow rather than via external workarounds. Using modern methods such as Vibe Coding, automations can even be defined via prompts, without the need for time-consuming development work. The competitive advantage lies not in the AI itself, but in the workflow built upon it – the editorial team’s unique way of working, translated into software. The editorial team is no longer merely a user, but becomes the architect of its own processes.
At the first user, technology media outlet Heise Online,the editorial team developed ‘NextWriter’, a tool based on InterfaceAI which integrates key workflows directly into InterRed.
Editor-in-chief Volker Zota says the in-house fact-checking process, previously time-consuming, reliant on experience and difficult to standardise, has proven particularly effective. “With NextWriter, it is now firmly embedded in the editorial workflow: at the touch of a button, the tool reads internal sources, checks statements for accuracy, compares in-house spelling conventions and takes editorial rules into account. The editorial team receives specific alerts regarding inconsistencies, missing sources or incorrect statements directly within the editor, along with concrete suggestions for corrections and source references.
“NextWriter is not an AI feature, but our editorial process within the system, and that makes a fundamental difference to everything we have tried with AI tools so far,” he says. “InterfaceAI forms the basis for further developing our editorial workflows in a secure and scalable manner, thereby embedding our own workflow logic within InterRed.”
At the heart of InterfaceAI lies its ability to systematically map editorial processes. This shifts the role of AI in publishing from a one-off support function to an infrastructural level of editorial work. InterfaceAI is based on three key principles: editorial teams can define their own individual processes and implement them directly within the CMS, ranging from simple automations to complex agent-based workflows. At the same time, InterfaceAI controls which actions agent-based AI is permitted to perform within the system, thereby ensuring reliable, traceable results. And finally, the system is not a black box: users can actively interact with the AI at any time, influence processes, intervene or adjust results. AI thus becomes an integral part of the editorial workflow, rather than an uncontrolled replacement for it.
Pictured (top) Heise headquarters in Hanover, and (below) Dr Volker Zota – “workflow instead of a standalone tool”