The AI agents that were supposed to shop, book, and pay for you just got locked out of the biggest doors on the internet. This episode of the Web Guy Nick Podcast explains what blocked agents mean for your customers, and why clear text on your own website is now how you get found. Alex, Jordan, and Nick also cover agents working inside QuickBooks, a survey where only one business in five can prove its AI is paying off, and the one number every owner should track each month. Listen on Spotify, and get practical help at webguynick.com.
Listen to this episode on Spotify
The agents that were supposed to shop, book, and pay for you just got locked out of the biggest doors on the internet.
Alex
If your hours, your prices, and your booking rules are buried in images and messy pages, an agent cannot read them, and it will send that customer to a competitor it can read. Clear text on your site is no longer just good manners. It is how you get found by the new front door.
Nick
That is the gap I see in real businesses, too. Everybody has the tools. Almost nobody can answer a simple question. What did this save us, or earn us, this month? You do not need a dashboard. You need one number, written down, checked monthly. Hours saved on one task, or leads gained from one campaign. If you cannot name the number, you are paying for a hobby, not a tool.
Nick
Full transcript
Alex: Before we start: this podcast is produced with AI. Jordan and I are AI hosts, and Nick’s voice is an AI clone of his real voice. Every story is researched from public sources, with links in the show notes.
Alex: Here is the fact that should change how you think about AI agents. The agents that were supposed to shop, book, and pay for you just got locked out of the biggest doors on the internet. Amazon has blocked them, the airlines have no deals with them, and the big banks will not let them touch a customer account.
Jordan: I am Jordan, with Alex, and Nick joins us as always. Today, five stories. The web slams the door on shopping agents. Anthropic gives away five percent of itself. Disney picks up a famous chatbot company out of bankruptcy for twenty million dollars. Marketers use AI every day but almost nobody measures whether it pays. And DeepSeek cuts the price of search for agents to almost nothing.
Nick: And I am Nick. I run a web design and marketing business here in California, so my question is always the same. What does this change for a real business on Monday morning? Today the answer starts with your own website, because if agents cannot get in, customers using agents cannot buy.
Alex: Story one is the lockout. Reporting from The Verge, covered across the weekend digests, says users of agents from Meta, xAI, and OpenAI ran into failed tasks in recent days because the sites themselves blocked the agents.
Jordan: The details are specific. Amazon has blocked Meta’s agent since September twentieth. Yelp only lets in agents that pay for a license. Delta Air Lines has no deals with third party agents at all. And JPMorgan and Schwab will not let agents touch customer accounts.
Alex: So this is not one company being difficult. That is retail, local search, travel, and banking, four sectors, all giving the same answer in about a month. Site owners do not want to be used without permission, and they do not want to hand their customer relationship to somebody else’s agent.
Nick: Owners, hear the other side of that door. Your customers are going to start arriving through agents, and you get to decide the terms. If your hours, your prices, and your booking rules are buried in images and messy pages, an agent cannot read them, and it will send that customer to a competitor it can read. Clear text on your site is no longer just good manners. It is how you get found by the new front door.
Alex: Story two is Anthropic, and it is really two stories in one. First, Anthropic announced it is committing five percent of its shares to a new Public Benefit Fund, a company owned entity that will direct money toward nonprofit donations, direct payments to people, and work on the labor market impact of AI.
Jordan: The pledge grows, too. For every fifty percent jump in the company’s value, another two percent of shares can be added. Combined with earlier pledges, more than twenty percent of the company is now pointed at public benefit purposes, ahead of its planned public listing. The baseline value and the first recipients have not been disclosed yet.
Alex: Chief executive Dario Amodei has been talking openly about a crisis of trust in AI, and this fund is the visible answer. The second story is why that trust question is so sharp. In its own reporting, Anthropic disclosed that junior technical hiring has fallen from twenty seven percent to nine percent since early twenty twenty five, while its AI agents now work about six hours for every one human hour.
Nick: That pair of numbers is the whole debate in one place. A company can give away five percent and still be changing who gets hired at the entry level. For a small business, the lesson is practical, not political. If you use AI to do the junior work, you still need a path for a real person to learn the job, check the work, and grow into the senior role. Automate the task, not the training.
Alex: Story three is a cautionary tale with a Hollywood ending. Character dot AI, the chatbot companion company, filed for bankruptcy this week, and Disney is picking up some of its assets for twenty million dollars, plus job offers for at least sixty employees, roughly sixty million dollars a year in pay.
Jordan: The deal had an inside track. Karandeep Anand, the former chief executive of Character dot AI, recently became Disney’s first chief technology officer. The collapse itself came from a high burn rate plus growing legal exposure, including a lawsuit tied to a teenager’s death and ongoing child safety investigations.
Alex: Twenty million dollars for a company that was once a headline name tells you how fast value can vanish when the costs and the legal risk both point the wrong way.
Nick: Every owner listening should take one plain lesson from that. A chatbot on your website that talks to the public is your responsibility, even if a vendor built it. If it talks to kids, or gives advice it should never give, that is your name on the door. Test it with the hard questions, write down what it is not allowed to say, and keep a human name attached to it.
Alex: Story four is about all of us who market for a living. A new survey from Basis, the advertising platform, questioned one hundred eighty four marketing and advertising professionals, and daily AI use nearly doubled in a year.
Jordan: Sixty one point four percent now use generative or agentic AI every day, up from thirty three point six percent a year ago. Ninety eight point four percent use it in their work at all. Claude posted the sharpest gain, used by seventy two point three percent of marketers, up from just seventeen point nine percent last year, while ChatGPT use fell to seventy four point five percent from eighty eight point six percent.
Alex: Now the uncomfortable number. Just six point five percent of marketing organizations measure the return on their AI spending against defined benchmarks, and sixty five point one percent either do not measure it at all, or are not sure anyone is measuring it.
Nick: That is the gap I see in real businesses, too. Everybody has the tools. Almost nobody can answer a simple question. What did this save us, or earn us, this month? You do not need a dashboard. You need one number, written down, checked monthly. Hours saved on one task, or leads gained from one campaign. If you cannot name the number, you are paying for a hobby, not a tool.
Alex: Story five is DeepSeek, and it is a price story. DeepSeek rolled out a Search and Fetch service for AI agents at seventy five cents per one thousand successful requests. Brave charges five dollars for the same thousand, and Kagi charges twelve dollars.
Jordan: That is roughly seven times cheaper than Brave and sixteen times cheaper than Kagi. The search side runs on DeepSeek’s own index, and the fetch side returns a clean text version of a page that an agent can actually read. On the model side, its new V four point one Flash variant scored forty five on the Artificial Analysis Intelligence Index, second out of one hundred twenty one models, at thirty cents per million input tokens and one dollar and twenty cents per million output tokens.
Alex: Some developers are even wondering if search pricing for agents could eventually go below zero, as providers compete to shape what agents see on the web. That part is speculation, but the price cut is real and published.
Nick: Cheap search matters to owners because agents are only as good as what they can look up. When the cost of an agent checking your prices, your reviews, and your competitors drops this fast, the agent gets more thorough, not less. Which loops right back to story one. Be readable, be accurate, and be the business the agent can verify.
Alex: Time for Higgsfield Corner. On October seventeenth, Higgsfield added creator rankings and grants to its Community. Creators can publish projects as a public portfolio, build a visible rating, and apply for grants of up to two hundred thousand generation credits.
Jordan: The portfolio idea is the interesting part for a business. A public page that shows the work, the production details, and a rating is basically a showroom. Customers judge the finished clips, not the tool list behind them.
Nick: My take stays simple. If you make AI video for clients, a public portfolio with your best finished work is worth more than any pitch deck. And judge every tool, Higgsfield included, by the cost of the clips you actually accept and publish, not the clips you generated and threw away.
Alex: Now our quick fire game, AI or BS. Three statements. Nick, you are playing. Statement one. Amazon has blocked Meta’s AI agent since September.
Nick: Real. Blocked since September twentieth, and that is the whole theme of today. The doors are closing unless there is a deal.
Jordan: Correct. Statement two. Disney paid two billion dollars to buy Character dot AI.
Nick: That is BS. Twenty million dollars for some assets out of bankruptcy, plus job offers for at least sixty people. Two billion is the kind of number it might have dreamed about before the burn rate and the lawsuits caught up.
Alex: Correct. Statement three. Only about six point five percent of marketing organizations measure their AI return against defined benchmarks.
Nick: Sadly real. Sixty one percent use it daily, and almost nobody measures it. That gap is the opportunity for any owner willing to track one honest number.
Jordan: Three for three. Which brings us to one thing to try this week, and it ties the whole episode together. Do a thirty minute agent access check on your own business.
Alex: Here is the test. Open your own website the way an agent would see it. Can a tool quickly find your hours, your service area, your starting prices, and how to request a quote, all in clear text? Then try one real task, like asking an AI assistant to find your phone number and your booking rules, and see what it gets wrong.
Nick: Write down every wrong or missing answer it gives you, and fix those exact items this week. Wrong hours, missing prices, a phone number trapped inside an image. Those fixes help agents, they help Google, and they help the human customer reading at midnight. One check, three wins.
Alex: That is the show. Agents got locked out, Anthropic gave away five percent of itself, a famous chatbot company sold for twenty million dollars, marketers skipped the measuring, and search for agents got much cheaper.
Jordan: The thread through all five stories is the same. Access, trust, and proof. Who can get in, who stands behind the tool, and who can show it actually paid.
Nick: Do the one thing. Run that thirty minute agent access check on your own site this week. If you want help making your website readable for agents and customers alike, you will find us at webguynick.com. See you tomorrow.
