A growing number of B2B buyers show up to a first sales call having already asked an AI assistant to explain the category, list the main vendors and summarize how they differ. Some arrive with a shortlist the assistant built. Some arrive with a misconception the assistant handed them. Either way, the first conversation no longer starts from zero, and sales teams that act as if it does lose ground in the first ten minutes.
Most writing about generative engine optimization focuses on marketing: content, schema, mentions. This piece is about the other half of go-to-market. What should sales and enablement teams do differently when the buyer’s research partner is a chatbot?
Find out what the buyer was told
The simplest change is to ask. Add one discovery question to the first call: “What have you read or asked so far, including any AI tools?” Many buyers will happily share the prompt they used and what came back. That answer tells a rep three things at once: which competitors are in frame, which criteria the buyer now believes matter, and whether anything about your company was described incorrectly.
Capture the answer in the CRM in a consistent field, not buried in call notes. Over a quarter, those entries become one of the most useful datasets a go-to-market team has: a record of how AI assistants are framing your category to real buyers.
Run the buyer’s prompts yourself
Enablement teams should maintain a short set of prompts that mirror how prospects actually research, by segment and by stage:
- Category prompts: “What is [category] and who are the main vendors?”
- Fit prompts: “Best [category] for a [size] company in [industry].”
- Comparison prompts: “[Your company] vs [competitor].”
- Objection prompts: “Is [your company] expensive?” or “Problems with [your company].”
Run them monthly across the assistants your buyers use and save the answers. Reps do not need to memorize them, but they should know the headline: where you appear, how you are described and what the assistant says about your weaknesses. A rep who already knows a buyer was told “strong product, steep learning curve” can address the learning curve before it becomes an unspoken reason to lose.
Build talk tracks for AI-shaped objections
AI summaries tend to compress nuance. A product that has a free tier and an enterprise plan might be summarized as “expensive.” A capability shipped six months ago might be missing because older sources dominate. A competitor’s framing of the category might become the default because they published more about it.
For each recurring theme in your prompt results and CRM notes, write a short talk track with three parts:
- Acknowledge what the buyer likely heard, without dismissing the tool they used.
- Give the accurate, specific version, with a public page the buyer can check.
- Offer proof that travels: a case study, documentation page or third-party review that says the same thing.
The second and third parts matter because the buyer will often go back to the assistant after the call. If your correction exists only in the rep’s words, the next answer will repeat the original error.
Close the loop with marketing
Sales hears the downstream effects of AI answers first. Marketing controls most of the upstream sources. The two need a direct line:
- A shared error log. When a rep hears something wrong about the company, it goes into one list with the prompt, the answer and the date, so marketing can trace and fix the source.
- Content requests from the field. If buyers keep asking assistants a question your site does not answer clearly, that is a page to write, in the buyer’s own words.
- Win and loss notes that mention AI. Tag deals where the buyer cited AI research, and look at whether those deals behave differently.
None of this requires new software. It requires agreeing on where the information goes and who reads it.
Arm reps with linkable answers
A follow-up email is now a source-seeding opportunity. When a rep answers a question about pricing structure, security, integrations or implementation, the best answer links to a public, crawlable page that states the same thing. That helps the buyer, gives them something to forward to colleagues on the buying committee, and reinforces the same facts AI systems can read. If a rep keeps attaching the same PDF because no public page exists, that is a gap worth flagging.
Be careful about what you promise
It is tempting to tell buyers that an AI tool is simply wrong. Sometimes it is. But when the summary reflects a real limitation, arguing makes the rep look less trustworthy than the chatbot. Concede real trade-offs plainly, explain who the product is and is not a fit for, and let the accurate picture do the work. Buyers notice honesty, and honest fit statements are also the kind of content that tends to hold up when assistants summarize you later.
Measure the right things
Useful indicators for a sales-side program are modest and concrete: the share of first calls where buyers mention AI research, the most common errors heard in the field and how quickly they disappear from prompt results after fixes, and win rates on deals where AI-shaped objections came up versus those where they did not. Track them for two or three quarters before drawing conclusions.
Teams that want help connecting the field signal to the content and source work behind it can look at a dedicated GEO partner. The underlying shift is the same either way: the buyer’s first briefing now comes from a machine, and the sales process has to start from what that briefing said.