On October 20, at its GO conference in New York, Demandbase launched three new AI agents. Two of them do the kind of work its customers already expect: build audiences, steer campaigns. The third is called the AEO Agent, and its whole job is to answer one question: what do ChatGPT-style assistants tell your buyers about you?
That a company built around account-based marketing now ships an answer-engine product says a lot about where B2B buying has gone.
What the AEO Agent does
According to Demandbase's announcement, the agent tracks how AI assistants describe and recommend a brand, down to individual members of a buying group. Marketers can compare those descriptions across "major AI assistants" (the release does not say which ones), check visibility on the questions that matter to them, and see which sources sit behind each answer.
From there it flags outdated information and content gaps, and hands off to Demandbase's other agents to run campaigns that fix them. It also claims to identify anonymous traffic that arrives from AI tools, at the account and contact level, so sales can follow up.
Demandbase says the agent draws on more than 342 million contacts across 13 million companies. Pricing and availability were not in the release.
"Buyers no longer start with a search"
The launch is framed around a term Demandbase is pushing hard: B2A, or business-to-agent marketing. The idea is that a growing share of vendor research is done by AI agents on a buyer's behalf, so a company has to be discoverable, understandable and credible to those agents, not only to people.
CEO Gabe Rogol put it bluntly: "buyers no longer start with a search." The company also leaned on a Gartner forecast that AI agents will intermediate 90% of B2B buying by 2028, about $15 trillion in enterprise spend.
You can argue with the 2028 number. It is harder to argue with the direction. When a buyer asks an assistant for the best tools in a category, the shortlist it returns is the shortlist. Vendors who are not named in that answer never get a demo request to lose.
Sources are the real product
The most useful line in the announcement is not the agent count. It is "identify the sources behind AI-generated answers."
That matches what anyone measuring this sees. Assistants rarely invent their opinion of a vendor. They assemble it from pages: review sites, comparison articles, Reddit threads, docs, and sometimes the vendor's own site. Semrush's 2026 AI Visibility Index, which covered 22 categories and four AI platforms, found that on average fewer than half of the 100 most-mentioned brands were also among the 100 most-cited domains. Being talked about and being the page the model reads are two different things.
So a visibility report that only says "you were mentioned 12% of the time" is half a diagnosis. The useful half is which pages shaped the answer, and whether any of them are yours.
What this means for smaller B2B teams
Demandbase sells to enterprise marketing departments with ABM budgets. Its customers will now get AEO readings inside a platform they already pay for. That raises the bar for everyone else in their categories, because the companies with the most resources are about to start optimizing for the same answers you want to appear in.
For a startup, the good news is that ChatGPT and Claude do not weigh ad spend. They weigh whether a clear, specific, crawlable page exists that answers the buyer's question, and whether third-party pages back it up. Three things are worth doing now:
- Ask the questions your buyers ask. Run your ten most important category and comparison questions through ChatGPT, Claude and Perplexity, and write down who gets named.
- Read the citations. For every answer that names a competitor and not you, open the pages it cites. That is your gap list.
- Fix the pages, then re-measure. Publish the page the answer is missing, correct outdated descriptions of your product, and check the same questions again a few weeks later.
The bigger picture
A year ago, answer engine optimization was a niche topic for SEO forums. This week, one of the largest B2B marketing platforms built a product around it and named the shift "business-to-agent." That is the signal: AI assistants are now treated as a buying channel with its own gatekeepers, and the vendors that get named are the ones whose pages the models trust.
You do not need an enterprise platform to see where you stand. You need the right questions, a careful read of what the assistants cite, and the patience to change pages and measure again.
ClawWorld runs AI visibility work for B2B startups. Start with a free baseline report: 10 buyer questions, who ChatGPT and Claude name instead of you, and the pages they cite.