Insights & articles

For years, people have debated whether AI systems may simply help themselves to the open web. Text, articles, images, knowledge – everything is collected, processed and replayed. Often without asking. Often without payment. Some call that innovation. Others call it plain free-riding.

Now there is at least a serious attempt to bring order: Really Simple Licensing 1.0, or RSL. The name is meant to deliver. RSL is meant to define how AI companies may use content from publishers and other creators – and under what conditions. That explicitly includes licence agreements.

Technically grounded Technically, the approach is surprisingly grounded. RSL is an open web standard, machine-readable, transparent and built on familiar mechanisms such as XML and RSS. So nothing entirely new, but an evolution of what the web has known for years.

More than yes or no The difference from current practice is substantial. The robots file – which is often ignored anyway – can only say yes or no at a very coarse level: indexing allowed or forbidden. RSL goes much further. It aims to provide a universal description of content rights and licence terms.

Site operators can specify that search engines may use content but AI search applications may not. Or that content may be read but not used to train AI models except for payment or other consideration.

Broad support The standard is already backed by organisations such as Cloudflare, Akamai, Creative Commons and the online advertising industry association. More than 1,500 media companies are said to support RSL.

Whether RSL really works Whether RSL ultimately works depends on final adoption. Chatbots must respect the standard. Crawlers must read it. And publishers must deploy it actively. But: it is a start.

For the first time, there is an attempt to govern the relationship between content creators and AI systems not through courts or bilateral contracts, but through a shared technical standard. Whether that tames free-riding for good remains to be seen. But one thing is clear: the open internet needs rules if it is to survive the AI era.

Have a good day – and always look forward to tomorrow.

Further reading: RSL 1.0 specification, Heise (German tech news): RSL 1.0 standard to govern use of content

I have led agencies for 25 years. In that time I have seen many crises – from the new economy to the pandemic. But AI tops everything. It changes our work and our clients' work fundamentally. That is why we are responding proactively and presenting a new service model.

Challenges for companies too Many of our communications clients face challenges: they know they must use AI to stay agile. But orientation is missing. Developments pile up. Every day brings a new model, 20 new tools and 50 alleged best practices. Add missing technical or prompt-engineering skills in the team, necessary process changes and the question of the right balance between AI and human creativity.

Service as a Software: the Kammann Rossi vision We engaged early, deeply and methodically with AI and LLMs. Our platform is called AssistantOS, and the new service model is called Service as a Software:

Human creativity We say clearly when AI hits its limits and classic creative work is needed.

Hybrid working model In projects we use AssistantOS together with our clients. Sometimes we provide the creative input, sometimes we execute more complex tasks. We fold new knowledge into new assistants.

Continuous optimisation We watch the AI market and always integrate the latest models and technologies into AssistantOS.

Consulting and training We train our clients' teams in working with AI and develop new use cases and assistants together.

Adaptation and training We adapt AI assistants to each client's specific needs and brand guidelines. That produces on-brand, high-quality content.

Benefits for both sides We give our clients flexibility between self-service and full-service, access to our agency know-how on demand, state-of-the-art AI technology without their own development costs, and more efficiency and creativity in day-to-day work.

For us as an agency, we can turn our unique knowledge and creativity into a licensable, scalable product.

We all talk about prompts and chat windows that have suddenly become intelligent. But honestly, I believe that is only a transition. The real future is not typing – it is speaking.

Audio solves old problems Audio conversations remove much of what frustrates people on the internet or intranet today: impersonality, irrelevance, complexity, information overload and fragmentation. A conversation delivers what systems could not before: it is dialogical, contextual, simple, focused – and it pulls everything together.

Marvin in beta For that we built Marvin, for example, as a beta. Marvin is the voice-first entry to our own AI environment AssistantOS – a voice bot that answers, asks follow-ups, executes actions and adapts to persona, prompt and context of each conversation.

The exciting part is not the technology And the exciting part is not the technology behind it. The exciting part is: Marvin changes how people interact with systems. Suddenly nobody needs to know where information lives, nobody needs to understand navigation, nobody needs to open menus or guess terms. You can simply speak – the way humans have communicated since we exist. And the system listens, understands and responds.

Voice as the biggest shift I am convinced: voice conversations on top of generative AI will be the biggest change in corporate communications since social media. Because voice lowers barriers. Because voice creates closeness. Because voice dissolves complexity.

Have a good day – and always look forward to tomorrow.

RAG stands for Retrieval-Augmented Generation. It describes a method where AI models are enriched with external knowledge.

The problem without RAG Large language models only know what was in their training. They know nothing about your company, your products or your internal processes.

The solution With RAG you upload your own documents. The system splits them into small chunks, builds semantic embeddings and searches them on every request.

Practical benefit An assistant with RAG can answer precise questions about your products, cite policies and work from your company knowledge – as if it had read every document.

I am always somewhere between nerd and curious. So one weekend I tried something: I built an AI-powered Minecraft bot called Freddi. On command he can mine, craft tools, build houses, fight monsters – all with local artificial intelligence.

How critical world knowledge is While testing and tuning, it became even clearer what I want to share: how critical world knowledge is – knowing the environment you act in. Freddi can only act if he understands what an axe is, what it is for, how to get wood and that without wood you cannot build a workbench. If he does not know he can dig, he will not get out of a hole either.

AI needs context That sounds trivial, but that is the point: AI needs context to act sensibly. Without knowledge of tools, materials and rules, the bot runs into walls or digs itself in. In the real world it is much the same.

Corporate communications need world knowledge When artificial intelligence is used in corporate communications, it must understand the world it works in. That means context about the brand, the company, language and more – and only we can really give it that. For example through system prompts that explain who it is, whom it speaks for and what style it should use – sober and precise like a press officer or warm and casual like a social media team.

Access to the right data It needs access to the right databases where company knowledge lives – products, projects, names, abbreviations, internal terms. Without those sources it cannot make real connections. It also needs world knowledge at a smaller scale: tone, audiences or company culture.

From bot talk to real corporate communications You can map that with knowledge collections or small brand memories that show the system how an organisation thinks and speaks. Only then does bot talk become real corporate communications. Only then does a model become an assistant.

That is the big difference between "having" artificial intelligence and applying it. An LLM knows billions of facts, but only context turns them into meaning. That is why the most important task today is to embed AI in a world it can understand – whether that is Minecraft or a global company.

Have a good day and always look forward to tomorrow.

Creative work and AI – that can sound like a contradiction. In practice, AI is a powerful tool for creative teams.

1. Brainstorming and ideation AI assistants can spar, develop ideas and bring new perspectives.

2. First drafts Whether text, image or presentation – AI can produce quick first versions to refine.

3. Variants and A/B tests Instead of one version, teams can generate dozens of variants and pick the best.

4. Translation and localisation Content can be adapted quickly for different markets and languages.

5. Reformatting and distribution From one blog post to a social thread, from a press release to a newsletter – AI helps reformat.

Does AI help in corporate communications – a topic debated as hotly as whether pineapple belongs on pizza. At Kammann Rossi we decided to act instead of debate endlessly. With our #keineagentur AI roadshow we want to show you how to use generative AI practically and effectively in your daily communications work.

#keineagentur? From an agency? Yes, you read that right. We at Kammann Rossi – in business for over 50 years – believe our role will change. The future of content creation increasingly sits with in-house teams, supported by smart AI tools.

What to expect In focused two-hour workshops we show you how to use generative AI – from language to image and video models – for your communications routine. Instead of PowerPoint slides we go hands-on in our secured AI environment AssistantOS, set up for the workshop. You can try everything in practice and get access to selected creative AI tools.

Concrete learning goals 1. How to integrate AI into your daily workflow 2. Which AI tools fit which communications tasks 3. How to weave company knowledge cleverly into your prompts 4. How to get AI to write in your personal style

Why join? - Networking snack after the workshop - Space for your questions - A look behind 50 years of agency experience

The GDPR sets clear requirements for handling personal data – including when using AI.

Where do the data live? What matters is where data are processed. With AssistantOS, all data stay in Germany on GDPR-compliant servers.

Multi-tenant isolation Every organisation must be strictly separated from others. Logical separation, access control and audit logging are mandatory.

No training on your data Important: your data must not be used to train AI models. With AssistantOS, that is guaranteed.

Documentation duties Companies must document which AI systems they use and how they work. A transparent architecture helps.

Corporate communications are at a turning point. Artificial intelligence lets teams work faster and more efficiently without sacrificing quality.

Three areas show the potential in particular:

1. Content creation Text, images and videos can be produced in a fraction of the time. The AI acts as a creative partner that delivers first drafts and refines ideas.

2. Personalisation Communications can now be tailored to different audiences without manually crafting every variant.

3. Knowledge management With RAG systems, teams can tap the full company knowledge base and fold it into their work.