Ask AI What Your Buyers Ask. Find Out If It Names You.

What You’ll Gain

• The new rule for getting found. Why answering specific questions now beats ranking for one big keyword.
• A GEO audit you can run yourself. Ask AI to audit your site, then ask it the questions your buyers ask and see who it names.
• A weekly routine that turns the audit into a scoreboard. Re-ask, collect new questions, build one page per gap, retest monthly.
• How two AIs share one set of instructions. Keep them in your website’s brain so Claude and Grokbot work from the same files.
• Where AI money went this week. $19.62B, a reported DeepSeek round, and chips and robots behind it.

Biggest Takeaway to Implement: This weekend, ask an AI what questions your buyers are asking about your kind of business, then ask those questions to ChatGPT, Claude and Grok and see whether they name you. Every question where they don’t is a page your site should build next.

0:00 SEO’s Golden Age Is Coming Back. Are You Ready?

Chris opened with a claim. For most of the last decade, the lesson of search was that organic results kept getting pushed down, the rules kept changing, and you had to buy ads to win. He said that is quietly reversing. The question is no longer how a search engine ranks you. It’s how the internet finds information about you when people have questions and you have answers. He compared it to 2008, when a good blog post could rank and bring in business right away. Businesses that make their sites visible to AI will be the ones AIs recommend. Olga welcomed viewers to a birthday edition of the show and set up the hour: the week’s news, then her playbook for getting found by AI, then funding.

4:14 The Company Whose AI Bill Beat Its Payroll

Anthropic filed to go public on October 26, listing on Nasdaq under the ticker ANT, and its filing opened the books. Olga asked Chris to guess how much of Anthropic’s own code Claude writes. The filing says: “Over 90% of the contributions to our codebase in September 2026 were written by Claude.” In April, Anthropic’s internal use of Claude, valued at list price, passed its monthly personnel expense. Revenue for the first half of 2026 was $16.29 billion, about 14 times the same period a year earlier, and the June quarter was its first with an operating profit, $974 million. Olga said that matters because the company needed to show it can make money, not just spend it. Chris guessed that much of Anthropic’s internal usage is inefficient and said that isn’t a big negative: there is room to get better. He said he is bullish. He uses the product every day, it works, and the company looks disciplined to him.

8:51 No More Coding Test. What Gets You Hired Now?

Felix Rieseberg, a hiring manager at Anthropic, wrote that he stopped asking candidates to do coding tests. That is his team’s practice, not company policy. His post’s title is the point: hunger beats tenure. Chris said it depends on the role. For extensions built on a system that already exists, he likes it. For foundational systems, where someone decides how the whole thing works, he called it a disaster, because it assumes the AI models do it right. A day later, GitHub’s Octoverse report said that as of September, developers start more than 40% of pull requests with an AI agent, up from 17% in March. Olga tied it back to last week’s story on Anthropic hiring fewer juniors: what counts now is how you orchestrate and solve problems. Chris added that knowing a language’s syntax is worthless knowledge. Understanding how systems are built is what lasts.

13:03 Microsoft Offered Buyouts. Would You Even Get the Offer?

Microsoft offered voluntary exit packages to about 60,000 employees, roughly 27% of its workforce, Reuters reported after Business Insider, and Microsoft confirmed it. The offer covers level 70 and below in the US, Canada, the UK and India, with decisions due in early December. These are buyouts, not layoffs. Olga said she reads it as a likely prelude to cuts, and turned it into a question for viewers: how do you make yourself so valuable that your company would never think to offer you one? Chris took the bigger view. The hiring boom of the pandemic years was the last burst of an old way of growing, and AI pushes in the opposite direction. He said that chapter is closed. The hosts then talked about revenue per employee as the metric to watch, and Chris asked for a scoreboard of it by company. Olga said she would build one.

17:58 OpenAI Tried to Cut Your Plan. Then It Doubled It.

OpenAI had planned to cut the usage included in its $200 Pro plan, reported as a 50% cut. After pushback, on October 29 OpenAI’s Tibo posted that it was doubling all usage limits on its Pro tiers and Business Premium instead, and that it had heard the “kind feedback.” BleepingComputer has the details. Olga joked that it was like telling a five-year-old “no” and then handing over two. Three days earlier, OpenAI released GPT-6.1 Terra, built to use a computer for you: clicking, typing and working across apps, priced at $1 per million input tokens. OpenAI says it costs about nine times less per task than Claude Opus 5.5, on its own test. Chris asked whether it’s better. Nobody knows yet. Then Google: Chrome’s autofill can now pull order numbers, tracking numbers and flight details from your Gmail into forms. It’s opt-in and US only. Olga pointed out that Claude in the browser can already do that.

19:58 Three Months of Paperwork in Four Hours. How to Check AI’s Work.

Olga asked Chris how long the first draft of a regulatory document for a new drug takes by hand. About three months, according to Anthropic’s October 27 post on drug development. At the biotech Enveda, Claude produced a first draft in four hours. In a batch-record review, it caught all 11 issues people had found, plus seven more they missed. The work is preclinical. Olga said this lines up with her own rule for trusting AI with a new task: do it yourself the first time and compare the results, to check it’s taking the right steps. Chris was more cautious. He said this is one of many painful steps in drug approval, and probably not the longest, even if AI handles that kind of rule-bound paperwork well. Both agreed paperwork is about to change.

22:58 Can You See What Your AI Actually Saves You?

On October 28 Anthropic added smart reports, surveys and pooled budgets to Claude Enterprise. Teams can see what people use Claude for, what it costs, where they get stuck, and what they get back. Olga called it a smart move to show return on AI, not just its cost. As a business owner, she said, she only cares about three things: did it save me money, did it save me time, and did it make me money. Chris added a fourth that people discount: fun and delight. Bringing something to life that used to be impossible is delightful, he said, and so are tools that make daily life better. Olga agreed. She has never worked harder or had more fun, and that set up the deep dive.

25:09 Anyone Can Vibe Code a Site. What Can Yours Do for You?

Olga framed the deep dive with a question she now asks every business owner she meets. Anyone can vibe code a pretty site. But what can it do for you? She doesn’t want to do the work that GEO audits hand back. She wants her AI to do it. Her demo used AI Business Plays, a site she built on PageMotor. Chris said what stood out to him watching her build it: she went straight to the customer, which here is the AI. When you ask an AI how to get cited, you’re asking the very thing that will read your site and decide whether to cite it. Any business owner who can ask the customer what they want should, he said. Olga asked the AI what it looks for, then a string of follow-up questions, and built the site and its content around the answers.

28:18 A Website That Publishes New Pages While You Sleep

Olga built AI Business Plays in three or four hours over two weeks by talking to Claude. It collects real AI plays from small business owners, each verified with its source, across many fields, because an idea from one industry can spark another. Her Grokbot saw the site had MCP, connected itself, and now adds new plays every day before she wakes up. Each play gets its own page: what the owner did, a link to the original post on X, step-by-step instructions to try it, and a prompt to run. Subscribers get the new plays by email every morning, and the forms and email come with PageMotor, with no outside tools. Chris broke down why it works. Each play page is specific to an industry and a kind of AI use, with the details an AI wants when someone asks a matching question. No human can keep up with that much specific content, he said. The AI has to run it, and a platform built for that removes the technical questions most owners can’t answer.

37:06 Two AIs, One Brain: Where Your Instructions Should Live

Olga showed her Claude project connected to the site. Claude and Grokbot have an inbox where they leave tasks for each other and scheduled check-ins during the day, so the work moves without her. She showed how she talks to the site in plain language, typos and all. The key, she said, is that nothing lives in Claude. Everything lives in the brain on her website. Chris explained why. Claude handles design changes, plugins and dashboards. Grokbot finds content, checks inboxes and does the managerial work. Both need the same instructions. Copy them into each tool and you maintain two versions, and when one falls out of date you get problems. The instructions need one home. The common answers are Dropbox, Google Workspace or Notion, which you don’t own. Your website is a space you do own.

41:00 Give Your Website a Job First

Olga opened her site’s admin dashboard on PageMotor: visitors today and over the last few days, with tracking turned on only about five days earlier. The first thing she did was give the site a goal, so it always shows her what she wants. Hers: 100 subscribers and three new PageMotor customers. The tracker said she was behind pace, which she said she didn’t care about yet. Once the site knew the goal, it told her what to work on first: getting found by AI. That’s where the GEO work started.

42:03 The GEO Audit Agencies Charge For, Fixed by Two Bots

Claude ran a GEO audit of her site, and Grokbot ran a separate one. You can simply ask your AI to evaluate your website for AI search. The result was a list of 26 items. Chris said agencies typically charge about $2,000 a month for this, with a three-month minimum and often a setup fee. Because the audit lives in the site’s brain, the site displays it as a table: what was wrong, and which AI is responsible for fixing it. The bots tried to assign some items to Olga, and she said no. She approves only what truly needs a human. Grokbot set up Google Search Console and Bing Webmaster Tools itself, which is why the two bots need a shared inbox. The fixes took two or three hours and were done by October 26. Two items are hers: getting listed on roundups of the best AI use cases, and adding the site to her profile on the Practical AI Show. Chris explained the brain again: the single source of truth for any AI that connects to the site.

46:59 Ask AI What Your Buyers Ask. Does It Name You?

The most important part of the audit was the questions. What Olga learned: it doesn’t matter which terms you rank for. What matters is which questions your buyers ask AI, and whether AI names you when they do. Grokbot has its own computer, so it asked other AIs 15 of those questions as if it were a buyer. The site came up on two, maybe on two, and not on 11. The two yeses were brand questions, which she said don’t matter much. The full recheck is set for November 24, when the bots rerun the same 15 questions. The questions are stored in the brain, every number has a source, and the schedule for the changes is set. Olga said she isn’t responsible for any of it.

49:31 Real Visitors or Bots? See Who’s Reading Your Site

Every morning at 6:30 the bots pull numbers from Google and Bing into one console on her dashboard: clicks, click rate and position, after about five days of data. PageMotor also counts visitors without Google Analytics and separates real people from bots. Olga cares most about the real people who stay. Chris said the bot traffic matters too: you want to understand the whole audience. The AI bots name themselves. ChatGPT sent the most and read many of her plays, Meta reads for training, Perplexity indexes the site, and Claude sent one. Chris said that tells you which plays AIs consider when answering users, and you can study those pages for a pattern and publish more like them. Kenn in the chat added a warning: make sure your web host isn’t blocking AI visitors. Chris said some hosts do, and called blocking AI bots insane for a business that wants to be found.

54:49 Who Shows Up Instead of You, and What to Do About It

Olga’s summary so far: list the questions your potential customers ask about your business, and if you don’t know them, ask AI to interview you and come up with them. She had Claude pull the 15 baseline questions from the brain, like “How can a small business use AI to save time? Real examples.” Each records whether the site was cited and who showed up instead. Because it knows who shows up, the AI can study what they publish and plan how to compete, then do the work. Chris made two points. First, SEO agencies rarely show you a table like this: they sign you to three months, report the good news, and you can’t see what they did. Here you see every to-do and its status. Second, search was always weak at long-tail queries, the very specific questions people actually have. AI is solving that. This is the age of specificity, he said, and the strategy is the opposite of chasing one big keyword.

1:00:28 From Audit to Scoreboard: A Weekly Routine You Can Copy

Chris described the day the audit list went from all red to all green while Olga worked in the kitchen. Olga said it was beautiful to watch the two bots work as a team: Grokbot has hands, and Claude builds and implements. Then: now what? The agents set a routine. Every Sunday they ask AI the buyer questions and track the results week over week. All week they add new questions from Search Console, Reddit, X replies and reader emails. On Sunday they bring Olga one or two questions where AI doesn’t name the site, she decides, and they build a page for each. Every month they retest every page they built to see if it started getting cited. Olga said she doesn’t know yet if it’s the right way, but she has a plan.

1:01:52 Live Build: Should a Small Business Use an AI Receptionist?

The most common question her buyers ask about AI in their business turned out to be about receptionists. Athena, her Claude, had already left Grokbot a task in the brain: prepare a brief. Grokbot works from X, so it can capture what people are saying about a topic right now, which Chris called underrated. Olga explained that she is still the “official dinger”: the bots check each other’s inboxes on a schedule, but when she needs something now, she types “ding.” On air she told Claude to build the page from the brief. It read the brief, a pricing file, the plays analysis and how the site’s existing pages are built, and Chris noted that the brain holds that design knowledge for any AI. A few minutes later the page went live at aibusinessplays.com/ai-receptionist. It opens with a short answer: yes, for the calls nobody picks up, like after-hours and overflow. It quotes owners who put AI on the phone in their own words, answers follow-up questions such as whether AI receptionists annoy older customers, and covers setup, cost and ready-made options. It links related plays from the site and cites outside sources. Chris called it a master class in interlinking that the best SEO agency wouldn’t match. The page is now tracked like the rest.

1:17:51 Copy This Play: How to Find Your Buyers’ Questions

Olga’s steps if you want to copy the play: figure out your goal, decide how often to check it, and talk to AI about how to get it done. Chris asked how she found the receptionist question. She asked Grokbot to search everywhere for the questions her potential customers ask about the topics her site covers, and Grokbot listed its sources as Search Console, Reddit, X replies and reader emails. For her sales training business, she said, she would give the AI her ideal client profile and her site, and have it check YouTube, Reddit, X and LinkedIn. The bots consolidated the results into the most common questions. Chris’s takeaway: the new SEO is specific. Claude built the page from known questions, linked other pages on the site, cited outside sources and answered sub-questions in one place. He predicted that a business whose site used to have 100 pages will probably have 1,500. Olga added a warning from audits she has run for friends: many sites are blocking AI and don’t know it. Her line on her bots: they do 95% of the work, and it’s not autopilot.

1:25:43 Where $19.6 Billion in AI Funding Went This Week

AI companies raised $19.62 billion across 85 companies from October 22 to 28. That was 61.3% of the $31.99 billion in venture funding, per Crunchbase. The biggest round was DeepSeek: Crunchbase records a $15.88 billion Series B out of Hangzhou. Reuters and Bloomberg reported about 100 billion yuan from anonymous sources, and DeepSeek hasn’t confirmed it. That round put 84.0% of AI dollars in China and 11.6% in the US. Olga noted that until now the giant rounds had gone to American companies. Chris’s list of the new top tier: Anthropic, OpenAI, xAI, SpaceX and DeepSeek. FuriosaAI raised $610.5 million out of Seoul for energy-efficient AI chips; the company says $620 million. Bedrock Robotics raised $600 million out of San Francisco for kits that make existing excavators autonomous (confirmed). Chris explained why retrofitting is the only approach that makes sense: heavy equipment passes from job to job for years, so you can’t wait for new machines. Pathos raised $237 million out of New York for agentic AI in cancer drug development, and Warp Sciences $215 million for a robot-run chemistry lab (both confirmed). Chris said a $215 million round used to be huge news and now ranks fifth. The full Week 48 report is here.

1:31:22 Track Why You Won the Deal, Not Just Why You Lost

Chris’s biggest takeaway: the move from the general to the specific. The long tail used to be the part of SEO everyone threw away. Now it’s where the money is, because users ask about their own situation, and the job of content now is to lay down a wide range of very specific answers. Olga’s came from her SalesUpLevel class, where she teaches sellers to sell more with AI. Sellers always log why they lost a deal. She challenged them to track why they won: why the customer came to them and what made them want to buy. Their AI now fills in a “why I won” field after every call. That, she said, is what GEO will thrive on: knowing why customers want you, and making sure your site answers it. Her ask for the weekend: talk to AI, see what people are asking, and make sure your site covers it.

Resources And Sources

• Ask AI What Your Buyers Ask. The slides as shown on air: the news and the GEO deep dive.
• AI Funding, Week 48. October 22 to 28: $19.62B across 85 AI companies.
• Anthropic: S-1 filing. October 26.
• Felix Rieseberg: Hunger beats tenure. His team’s hiring practice.
• GitHub Octoverse 2026. Agent-started pull requests.
• Reuters on the Microsoft buyouts. October 26.
• BleepingComputer on OpenAI’s Pro limits. October 29.
• OpenAI: GPT-6.1 Terra. October 26.
• Google: Chrome autofill from Gmail. October 28.
• Anthropic: early progress in drug development. October 27.
• Claude Enterprise: smart reports, surveys and pooled budgets. October 28.
• AI Business Plays. The site from the deep dive.
• Should a small service business use an AI receptionist? The page built live on the show.
• PageMotor. The platform the site runs on.
• Weekly AI Funding Tracker. 48 weeks.
• Episode 62. The week before, including Anthropic hiring fewer juniors.
• The email list. The notes after every show.