AI + Content
Your Buyer's First Impression Now Happens Inside an AI Agent. Here's What a SaaS Founder Must Publish to Get Recommended
Dan McDermott//May 22, 2026//7 min read
A few months ago, I started noticing a pattern in sales conversations. Prospects were showing up already oriented. They had opinions about pricing, comparisons in mind, sometimes even objections ready before I'd said anything. They weren't coming in cold. They were coming in shaped.
I started asking: where did you start your research? The answer, more often than not, was ChatGPT, Perplexity, or Google's AI summary at the top of the SERP. They'd described their problem to an AI, gotten a shortlist back, and then come to me to validate it. The evaluation had already happened. I just wasn't in the room for it.
That's the moment this got real for me. Not a report, not a conference talk. There's now a sales call happening before the sales call, and most founders have zero visibility into it and zero strategy for it.
G2's 2026 Answer Economy report surveyed 1,076 B2B software buyers in March 2026. 51% said they now start their research in an AI chatbot more often than Google. Twelve months earlier that number was 29%. That's not a gradual drift. That's a full category flip inside twelve months. More striking: 69% of buyers chose a different vendor than they originally planned based on AI guidance, and one in three bought from a brand they'd never heard of before the AI surfaced it.
Read that last number again. A third of buyers are purchasing from companies they didn't know existed until an AI named them.
Gartner's strategic predictions put the long arc in focus: by 2028, 90% of B2B buying will be AI-agent intermediated, routing over $15 trillion in B2B spend through AI agent exchanges. (Independently reported by Digital Commerce360.) At Google I/O on May 20, 2026, Google announced Information Agents, background agents that scan blogs, news, and social content 24/7 on a user's behalf before they ever visit a vendor site.
This is not a future trend to monitor. It's the current condition.
For a seed-stage founder with no established category authority, there's actually an argument for optimism here. The shortlist isn't locked. In most software categories, there's no LLM incumbent yet. The brands showing up consistently aren't necessarily the biggest. They're the ones that gave the AI something useful to work with.
Here's the uncomfortable part. Ranking on Google doesn't help you here. Most of what you've published isn't what AI agents pull from.
The Foundation x AirOps study of 57.2 million LLM citations across 50 brands and 7 verticals found that brand-owned content accounts for just 10% of what LLMs cite. At unbranded, category-level queries. Exactly the moment a buyer is building a shortlist. That drops to 2.2%. In 85% of responses to those queries, no brand-owned source is cited at all.
So when a buyer types "what's the best [your category] tool for [your use case]," your homepage and blog posts are almost certainly not being pulled.
The other failure mode is vague positioning. "Flexible." "Powerful." "Built for modern teams." That's in every deck in every category. An AI can't build a shortlist out of that. It needs to answer: what does this do, for whom, with what result. If your content doesn't answer those questions in plain, structured language, you don't make the shortlist. There's no page two in an AI response.
Most AI SEO content right now is written by people claiming authority they don't have.
Most founders have no idea how LLM citation actually works at a technical level. Honestly, neither do I. Even people inside Anthropic and Google are still working this out. The signals are noisy, the models change, and anyone selling a definitive framework for "ranking in AI" is probably selling confidence more than insight.
But here's what is clear: good content that answers specific questions, written plainly and structured well, is what wins. Not keyword density. Not technical schema tricks. The principle underneath all of it is the same one that made Google work before it got gamed: be the best answer to a real question.
Resist the complexity theater. If someone's selling you a seven-step LLM ranking framework, close the tab. Write the one article that answers the most common question your buyer actually types.
If brand-owned content accounts for only 10% of LLM citations, where's the other 90% coming from?
Foundation x AirOps broke down the third-party sources: Reddit accounts for 20.8% of all external citations in LLM responses across B2B SaaS. It's the single largest source in 6 of 7 verticals studied. YouTube is at 13%, LinkedIn at 11%, review sites like G2 at 4%.
Your blog is not the primary input. Reddit threads, YouTube walkthroughs, LinkedIn posts from practitioners, and G2 reviews are doing more work for your brand inside AI responses than anything on your own domain.
The conclusion is not "don't bother with your site." It's the opposite. Your owned content sets the facts. Third-party sources need something accurate to build on. If a Reddit thread about your category gets things wrong about your product, the AI cites that. If your pricing page is clear and your comparison content is honest and specific, it becomes the factual foundation that other sources reference.
The practical priority order for a one-person marketing team:
First, get your own house in order: named customer outcomes, specific use-case pages, honest comparison content, clear pricing.
Then earn the external: LinkedIn posts, community threads, real G2 reviews.
Start with owned. Earn the external.
Do one thing before anything else: search for yourself. Right now. Across every engine and every AI.
Here's the exact sequence.
First, write down 5 to 10 prompts a buyer would actually type about your category and the problem you solve. Not your brand name. The problem. "What's the best tool for [specific use case]." "How do I handle [specific pain point]." "Compare [your category] options for [specific buyer type]."
Second, run every prompt across ChatGPT, Perplexity, Google AI Mode, and Claude. All four. The responses differ more than most people expect.
Third, note what comes back. Who gets named, who doesn't. What's said about the named brands. Whether you appear at all, and if so, whether what's attributed to you is accurate.
Fourth, identify one specific content gap to close this week. Not a content strategy. Not a six-month plan. One gap. One answer you're not currently giving that your buyer is asking.
This is a weekly discipline, not a quarterly review. The AI responses are not static. New content gets ingested, sources shift, and the brands showing up in week one of a category aren't guaranteed to be there in week eight.
If you run the audit and find you're invisible, that tells you exactly where to start. Not with a rebrand. Not with a new website. With one specific, well-written answer to a question your buyer is actually asking.
The sales call before the sales call is already happening. The only question is whether you're in the room for it. If you want to run through this audit together and find where the gaps are, let's talk.