Found in Google Search
The foundation. If a page is not in the index, no agent will find it either. In its own guide of May 15, 2026, Google writes that optimizing for generative AI features in Search is still SEO.
Ask ChatGPT, Gemini or Perplexity for a suitable company and you get a short list with reasons instead of a page of results. Agentic marketing means preparing your offer so that an AI agent can find it, understand it and verify it. SEOX remains your SEO and GEO agency. Agentic marketing is an additional area that builds on that work.
Agentic marketing is marketing for customers who let an AI agent do the searching. Acting for a person, the agent gathers providers, matches them against the request, checks their claims and suggests a shortlist. The goal is for your offer to appear in that process and to be described correctly.
The foundation. If a page is not in the index, no agent will find it either. In its own guide of May 15, 2026, Google writes that optimizing for generative AI features in Search is still SEO.
ChatGPT, Gemini and Perplexity mention your brand when someone asks about your topic. We build this AI visibility under GEO and LLMO for AI answers.
The agent matches your offer against a specific request, checks your claims and finds the way to send an inquiry or make a booking.
A customer describes a situation, for example: “I need someone by Friday to check the heat pump in an old building, budget 500 francs.” We describe the path to the shortlist in seven steps.
This is our working model, derived from how the systems behave, not an official statement by the providers. It is explained in detail in our guide AI agents for businesses.
The situation becomes sub-questions: service, location, deadline, price.
The company loses if it does not know these questions its customers ask.
The agent collects candidates from the search index, business profiles, feeds and websites.
The company loses if it does not appear cleanly in any of these sources.
For each candidate, it checks whether the offer fits every sub-question.
The company loses if information is missing or too vague.
It reads third-party sources: reviews, media, directories.
The company loses if these say nothing or say something different.
It looks for evidence behind claims, such as certificates, references or prices.
The company loses if its claims stand without proof.
It names a small number of companies and gives reasons.
The company loses if a competitor is better documented in steps 3 to 5.
A link, a call, an inquiry, and later the booking.
The company loses if the way to an inquiry is long or an agent cannot use it.
There are many forecasts about AI agents and few measurements. We rely on sources you can read for yourself and give the date and origin of the data for each one.
| Finding | What the source says | Source |
|---|---|---|
| Each system uses different sources | For local businesses, Google most often cites the Google Business Profile (28.6 percent of citations), while ChatGPT most often cites Yelp. Around 93 percent of the cited domains are websites of individual businesses. | BrightLocal, August 26, 2026, data from the US, the UK and Australia |
| Unproven claims cost deals | According to a Forrester prediction, two thirds of B2B providers will lose deals because buyers cannot verify their claims. | Forrester, October 21, 2026 |
| ChatGPT introduces business profiles | Since November 10, 2026, companies can claim and maintain a profile in ChatGPT, for now as a beta. Local shops and restaurants can currently only claim an existing listing in the US. | OpenAI and OpenAI help page |
It starts with the Agent Check. It shows at which of the seven steps you lose out today. That tells us what comes first in the project and what we measure on an ongoing basis afterwards.
The starting point, one-off. We ask 20 questions the way your customers would ask an AI agent and run them through several AI systems. We also briefly review steps 2 to 5.
The project. Together with you, we record what an agent needs to know about your offer: exactly what you do and for whom, price range, service area, lead times, conditions and what you do not do. We then roll this information out to your website, structured data and profiles.
Monthly. The question catalog runs through several AI systems at regular intervals. We see whether you are mentioned, who else is mentioned, which sources the system cites and whether what it says about you is correct. We trace wrong statements back to their source and have them corrected there.
Three parts of this area have a page of their own. The Agent Check shows which of them matters most for you first.
We check whether an agent can read and use your website, set up an Agent Card based on the open A2A protocol and add structured data that agents can read.
Agent readiness for your companyWe describe your offer as a machine-readable catalog so that the directories and catalogs agents query can include you.
Discovery optimizationFor shops: product data and feeds for ChatGPT and Gemini, and your own shop chatbot that leads to checkout. The page describes the current situation in Switzerland and our services.
Agentic commerce for Swiss shopsThe field is young. We tell you up front what you can expect, so you can judge the work correctly.
Today AI systems mainly shape searching and comparing; hardly anyone buys through them (Forrester, May 28, 2026). We do not promise predictable revenue from this channel.
Which company an agent suggests is decided by the provider of the system. In its terms for business profiles, OpenAI rules out guarantees of placement, visibility, traffic and revenue. We deliver the groundwork and the measurement.
What works on Google does not work the same way on ChatGPT, as BrightLocal's analysis of sources shows. That is why we measure each system separately.
The platforms do not release real user queries. We measure with a fixed question catalog. What can be shown is the trend over time, rarely the effect of a single measure.
Agentic marketing is marketing for customers who let an AI agent do the searching. Acting for a person, the agent gathers providers, matches them against the request, checks their claims and suggests a shortlist. The goal is for your offer to appear in that process and to be described correctly.
SEO makes sure your pages are found in Google Search. GEO makes sure AI systems such as ChatGPT mention your brand in their answers. Agentic marketing goes one step further: an agent should be able to match your offer against a specific request, check your claims and find the way to send an inquiry.
No. Agentic marketing builds on SEO and GEO. An agent only finds what is in the index and appears in sources. In its own guide of May 15, 2026, Google writes that optimizing for generative AI features in Search is still SEO. SEOX remains your SEO and GEO agency; agentic marketing is an additional area.
With the Agent Check. We ask 20 questions the way your customers would ask an AI agent, run them through several AI systems and briefly check whether you are found, whether your offer fits the request, whether third-party sources say the same and whether your claims are backed up. The result is a traffic-light rating for each step.
No, nobody can. Which company an agent suggests is decided by the provider of the system. In its terms for business profiles in ChatGPT, OpenAI rules out guarantees of placement, visibility, traffic and revenue. We put the groundwork in place and measure how your mentions develop.
As preparation, yes; as a revenue channel, not yet. Today people mainly use AI systems to search and compare, and they rarely buy or book through the agent. The work on clear information and evidence stays with you: afterwards it is also there for visitors to your website.
Agentic commerce is the part for online retailers: product data and feeds for ChatGPT and Gemini, and a shop chatbot that leads to checkout. Agentic marketing is the umbrella term and also applies to service providers who want to receive inquiries or bookings.