If a buyer asks an AI assistant which vendors to consider, the shortlist can form before your website is ever visited. G2’s April 2026 survey of 1,076 B2B software buyers found that 51% now start research with an AI chatbot more often than with Google. That changes what “being found” means for a services or software business selling to other companies.
- G2’s March 2026 survey of 1,076 B2B software buyers reports 51% start with an AI chatbot more often than Google, up from 29% in 2025.
- In the same survey, 69% chose a different vendor than first planned because a chatbot recommended it. Buyers also trust review-site citations and consistent answers across chatbots.
- Gartner’s forecast that 90% of B2B buying will be AI agent intermediated by 2028 is a prediction, not a measurement, and no methodology accompanies it.
- Demandbase launched an AEO Agent on October 20, 2026 to monitor how AI assistants describe a company. Pricing and availability were not stated.
- Start by checking how assistants describe you today, then fix the sources they draw on.
What the G2 survey measured
The strongest evidence on buyer behavior comes from software review platform G2. Its research article “The Answer Economy,” published April 15, 2026, reports a March 2026 survey of 1,076 B2B software buyers and decision-makers. According to G2, 51% now start research with an AI chatbot more often than with Google, compared with 29% in 2025.
The same survey reports that AI chatbots are the top source influencing buyer shortlists, ahead of review sites, analyst firms and vendor websites. G2 also found that 69% of respondents chose a different vendor than they first planned because it appeared in a chatbot’s recommendation.
Two caveats apply. G2 sells software-buyer data and sells to vendors, so the framing favors its own relevance. And the respondents are software buyers, so the figures may not transfer to every services purchase. The direction is still clear: for software, an AI answer now sits upstream of the vendor website.
What Gartner’s forecast does and does not show
Gartner’s October 2025 predictions for 2026 state that by 2028, 90% of B2B buying will be AI agent intermediated, pushing more than $15 trillion of spend through agent exchanges. We could not open Gartner’s release directly, so we rely on the wording Demandbase quoted from it and on the release date it linked.
The forecast describes intermediation, meaning an agent sits somewhere in the purchase. It does not say agents will make purchase decisions unaided, and we found no published methodology behind the 90% figure. Treat it as a planning scenario, not a measured adoption rate.
The implication for marketers is practical. Whether the 90% arrives by 2028 or later, G2’s survey shows the research stage already runs through assistants. Planning around that is cheaper than waiting for the forecast to be proven.
What Demandbase launched
On October 20, 2026, Demandbase announced three agents at its GO conference in New York. The AEO Agent: it compares how major AI assistants describe a company to each member of a buying group, shows visibility across relevant questions, and identifies the sources behind AI-generated answers. A Campaign Agent and an Audience Agent were announced alongside it.
AEO: Answer Engine Optimization, the practice of making a brand’s information easy for AI assistants to retrieve and state accurately. It overlaps with GEO, which we compare in GEO, AEO and SEO.
The release states that the agents are available through APIs and Model Context Protocol connections. It gives no general availability date and no pricing, and it offers no performance results. That makes it a signal of where vendors are heading, not evidence that the approach works.
Evidence compared
The sources differ in how much weight they can carry. The table separates measurement from prediction and from product announcement.
| Source | Type | What it supports | Limit |
|---|---|---|---|
| G2, April 2026 (n=1,076) | Buyer survey | AI chatbots are a common starting point and shortlist influence for software buyers | Self-reported; software purchases only |
| Gartner, October 2025 | Forecast | Analysts expect agents to enter B2B purchasing by 2028 | No published methodology found |
| Demandbase, October 2026 | Product launch | Vendors are building tools to monitor AI descriptions of brands | No pricing, availability date or results |
What to check first
G2 reports that buyers treat citations from review sites as their top trust signal and read consistency across chatbots as a sign an answer is reliable. When answers conflict with what they know, they dig further. That suggests the sources behind the answer matter as much as your own pages.
- Ask ChatGPT, Perplexity and Google’s AI Mode the questions your buyers ask, and record how each describes you.
- Note which sources each answer cites, and whether your review-site profiles are current.
- Correct outdated facts at their source, starting with your own site and the review profiles cited most.
- Repeat monthly, since answers vary run to run, as we explain in how to measure AI search visibility.
Third-party mentions carry weight here too, which is the subject of our analysis of brand mentions versus backlinks.
Frequently Asked Questions
Do most B2B buyers now start research with AI?
G2’s March 2026 survey of 1,076 software buyers reports 51% start with an AI chatbot more often than Google. That covers software purchases and is self-reported, so it does not describe every B2B category.
Will AI agents handle 90% of B2B buying by 2028?
Gartner predicts 90% of B2B buying will be AI agent intermediated by 2028. It is a forecast without a published methodology, and intermediation does not mean agents buy without human approval.
Do I need an AEO tool like Demandbase’s AEO Agent?
Not necessarily. Demandbase has not published pricing or results. A manual monthly check of how assistants describe you, and which sources they cite, gives you the same starting information at no cost.
If buyers are forming shortlists inside AI assistants, the work is part content, part reputation and part measurement. At Grow with Hafsa we help US and UK companies audit how AI assistants describe them and turn the findings into AI search and performance marketing priorities. Our guide to evaluating an agentic marketing vendor is a useful companion before you buy any monitoring tool.
Shortlists built by AI often draw on review profiles. Our post on reviews and AI search visibility under FTC and UK rules covers how to build them lawfully.