Media Informmation, 23 July 2026
Adaptive AI agents
Whitepaper by ERGO and Google on self‑evolving agents: how companies can preserve and use their know‑how for the future.
A new whitepaper by ERGO in cooperation with Google Cloud shows how self‑evolving agents can capture and preserve implicit in-house knowledge, and harness it for the entire organisation long-term.
The ERGO Innovation Lab, an incubator for new ideas and spin-ins within the ERGO Group, has published a new report in cooperation with Google Cloud. Entitled “From Tribal Knowledge to Self-Evolving Agents – Capturing, Governing, and Scaling Institutional Knowledge in Financial Enterprises,” the report examines how companies – and particularly insurers – can, with the help of adaptive AI agents, capture, preserve and scale what has often been only undocumented experiential know-how (“tribal knowledge”). The fact is that a significant proportion of business-related knowledge remains, to this day, neither formally documented nor systematically accessible. Instead, it stays embedded in routines, situational decisions, or within the staff members themselves. And demographic change carries the risk that this knowledge will thus increasingly be lost to organisations. According to figures from the German Insurance Association (GDV, German only), in Germany a full quarter of those employed in the insurance sector belong to the baby boomer generation. And so a large share of this very skilled workforce will be retiring over the next five years – and taking their wealth of experience with them.
“Self-evolving” or “self-improving” agents could help to preserve and make this knowledge accessible in the long term. These agents are adaptive AI systems that not only carry out tasks, but can also review their own strategies, workflows and, in some cases, even their own architecture over time, and adapt and improve them where necessary. The report includes a concrete implementation example of an “AI content engine”, as well as further suggestions of where such agents could be deployed across insurers’ value chain in future.
The next stage in the development of agentic AI involves not only greater automation, but also how companies preserve, develop and harness knowledge. In view of demographic change and the worsening shortage of skilled workers, this is becoming increasingly important. Indeed, without a sustainable strategy, organisations run the risk of losing their decision-making criteria, contextual knowledge, and practical experience. The value of self-evolving agents lies where traditional automation ends: in experience, context and judgement. That is precisely where they can help to systematically capture individual expertise across generations, and transform it into scalable institutional knowledge.
In highly regulated industries like insurance, the true power of agentic AI lies in human-guided autonomy. By pairing Google Cloud’s secure, scalable AI ecosystem with ERGO’s deep domain expertise, this whitepaper provides a forward-looking blueprint for preserving and scaling institutional knowledge responsibly.
With regard to ensuring that self-evolving or self-improving agents can be safely and compliantly deployed within the highly regulated financial sector, the whitepaper also examines the requirements for what is known as “enterprise readiness”. This includes, for example, clear governance and compliance structures, identity and access management, observability concepts, data protection measures, and active change management.
Furthermore, the publication outlines the prospects for further development: from initial, tightly controlled agents towards collective learning systems that share knowledge and themselves generate new learning opportunities – for example, through simulations and synthetic cases. The central question in this regard is how the relationship between human expertise, in-house knowledge and adaptive AI systems can be shaped in the long term.
The ERGO Innovation Lab also has other publications on generative AI and large language models for the insurance sector, such as “The Next Generation of Generative Artificial Intelligence: Advances and Multimodal Applications in the Insurance Sector” and “ChatGPT and Large Language Models: An Introduction with a Focus on the Insurance Industry”. Both publications are also available free of charge for download on the whitepaper focus page.
This media information contains forward-looking statements that are based on current assumptions and forecasts of the management of ERGO Group. Known and unknown risks, uncertainties and other factors could lead to material differences between the forward-looking statements given here and the actual development, in particular the results, financial situation and performance of our Company. The Company assumes no liability to update these forward-looking statements or to conform them to future events or developments.
