IC 2012 from Stuttgart to Düsseldorf - an emergency stop just before Cologne, the smell of hot brake pads, continuing at walking pace and a 90-minute delay.
This is starting well, I thought.
But the following two days at the OMT Summit 2026 made up for everything.
2500 participants and 140 top speakers!
Here is a summary of my highlights and learnings.

The paradigm shift in the agentic web

The search landscape in 2026 has passed the turning point. As Florian and Alex analyzed at the OMT Summit, the "state of search" is changing fundamentally: away from the classic list of ten blue links and towards direct answers (zero-click searches) and dialog-oriented recommendations by AI agents.

While Google continues to dominate the search volume with a market share of around 90%, the dynamics of language models (LLMs) can no longer be ignored: LLMs are expected to claim 25% of the market share by the end of the year. Florian Ries (enyma) puts forward the core strategic thesis: visibility in AI is not statistical noise, but the result of clear trust signals. In the age of the "agentic web", recommendability is decisive for business success.

E-E-A-T and digital authority in the age of AI

The E-E-A-T concept (Experience, Expertise, Authoritativeness, Trust) is undergoing a massive upgrade thanks to Generative Engine Optimization (GEO).

Experience: "Genuine human experience" is the only asset that AI cannot imitate, according to Florian. AI merely "hallucinates" experience. Strategically, this means: Focus on original data sets, e.g. own market studies, expert interviews and user-generated content, which act as "first-hand experience".
Expertise & topical authority: Away from isolated content, towards networked topic worlds. The author entity becomes the decisive ranking factor. An expert must be anchored in the network via external references (interviews in specialist media, guest articles, links to associations) in order to be considered a trustworthy signal for LLMs.
Authoritativeness (authority): There is a huge gap here between self-perception and external perception. Data shows: 85% of the information that AI outputs about a brand comes from third-party sites. Only 15 % is obtained directly from the brand website. Digital authority is therefore primarily created through a strong external footprint (PR, specialist articles, citations).
Trust: In addition to technical signals such as imprint and HTTPS, AI evaluates the consistency of the brand across all channels.

Technical optimization for LLMs

In order for language models to process content efficiently, websites must be designed to be technically "LLM-friendly".

How crawlers see: Stefan Kiecker (SEOCATION) illustrates the problem of JavaScript rendering with a comparison: to humans, a website is like a Lego-built Homer Simpson. From a distance, the figure is immediately recognizable. For the crawler, however, it only consists of individual parts. If the "bricks" (modular data) are not immediately accessible due to complex JavaScript or asynchronous database queries, the AI only sees "a hedge" - the page appears empty and information is lost.
Energy efficiency as a ranking factor: LLM crawlers act economically ("lazy crawlers"). A Google search consumes approx. 0.3 Wh, a RAG query (AI search with web access) already 15+ Wh and image generation up to 100 Wh. Since computing power is expensive, LLMs prefer static, lean pages that can be processed without massive JS rendering.
NLP before schema markup: While structured data (Schema.org) continues to make sense, LLMs now understand natural language (Natural Language Processing) like adults. Subject-predicate-object structures are more important for entity recognition than any JSON-LD file.

Content strategy: From keywords to entities and context

GEO requires optimization for citation and brand context.

Citability: Olaf Kopp uses scores to evaluate content. The decisive factor is not the word count, but consistency and specificity. Holistic 5,000-word monsters are dying out; precise, modular units are in demand.
Brand context optimization: Brands need to be positioned through "co-occurrences". The AI must consistently link the brand name with attributes such as "sustainable" or "B2B specialist".
Grounding Pages: Judy Lewis (OVB) and Tobias Fischer (add2) implement these "Wikipedia pages for AI". Grounding pages provide purely factual, modularly prepared data about the company that is easy for AI agents to extract.

Risk management: the "support trap" and legal protection

Lawyer Chrisopher Unseld warns of the existential danger of platform dependency (Google, Meta, Amazon).

The critical 4-week window: In the event of an algorithmic account suspension or abrupt policy change, exactly one month remains for the urgent legal protection. If you miss this deadline, you could face years of legal disputes.
The support trap: Don't fall into the trap of endless support ticket loops. The 4-week period for legal action begins the moment you learn of the block - regardless of whether you are still chatting with a support bot.

Summary and action plan 2026

The transition from SEO to GEO means the end of pure keyword hunting. The new currency is Digital Authority.

Measures for 2026:

Maximize external footprint: Invest in PR and expert profiles on third-party sites, as 85% of source citations in AI results link to them.
Think in entities: Modularize your content and use structures for maximum "LLM-ability".
Technical homework: Reduce JS ballast and optimize for energy-efficient crawlers.
Brand strategy as a foundation: Use grounding pages and brand identity blocks for consistent signals.

Olaf Kopp from the agency Aufgesang concludes with an urgent appeal: "Do your homework on your brand strategy before you optimize for GEO." If you don't have a clear identity, you will remain invisible to the agents of the future.