In partnership with
The big AI companies spent this week deciding two things that touch you directly: whether they can train on your website, and how cheap their models get. We read the filing, the standard and the price sheets so you do not have to. Figures from companies are labeled as such.
285 Free Prompts to Save Time
Your inbox isn't the problem. Neither are your meetings. It's the hours you spend on work AI could do in minutes.
285 copy-paste prompts: built for email, meetings, admin, and delegation. Each one has a clear role, a specific task, and a defined output.
No setup. No prompt engineering. Just copy, paste, and get your time back. Free when you subscribe to Mindstream.
TODAY IN AI
3 things that happened while you were busy
1. Google, OpenAI and Microsoft backed one way to keep AI off your site.
Anthropic proposed a single robots.txt rule, a token called __usage_ai-training, that lets a website say in one line whether its pages can be used to train AI, instead of writing a separate rule for each company. Google, OpenAI and Microsoft announced support the same day the draft was published. Read TNW's report. It is a preference, not a lock, and the companies will not honor it until 2027, but it is the clearest opt-out site owners have had.
2. Anthropic filed to go public.
Anthropic, the maker of Claude, filed its S-1 to list on Nasdaq under the ticker ANT, reporting $16.3 billion in revenue for the first half of 2026, up from about $1.2 billion a year earlier, alongside a large net loss driven mostly by a non-cash charge. Those are the filing's figures. See CNBC's coverage. The share price is not set yet, and a listing is reported for mid-November but unconfirmed.
3. OpenAI cut the price of its workhorse model in half.
OpenAI released GPT-6.1 Terra, a mid-sized model for everyday computer tasks, at $1 per million input tokens and $5 per million output, half the price of the Sol model it shipped weeks earlier, with faster agent runs. OpenAI's own benchmarks show large gains over the previous Terra. Read the details. Cheaper tokens mean the automations you build cost less to run.
TODAY'S TUTORIAL
Tell AI crawlers whether they can train on your site
The new opt-out lands in 2027, but you can set your preference today so it is ready when the crawlers start honoring it. Here is how to add it to your site's robots.txt, which works for a Shopify store, a blog or a newsletter site. See the standard.
1. Find your robots.txt. It sits at yoursite.com/robots.txt. On Shopify you edit it through the theme code under robots.txt.liquid; most site builders have a settings field for it.
2. Decide what to protect. Choose whether you want to block training on the whole site or just some sections, like paid guides or original photos, while leaving your public pages open.
3. Add the rule. Create a group for the new token: a line reading "User-agent: __usage_ai-training" followed by "Disallow: /" to block everything, or specific paths like "Disallow: /guides/" to block only those.
4. Leave a door open if you want reach. You can allow one folder, such as a press or about page, so AI answers can still cite you while the rest stays off limits.
5. Save and check. Reload yoursite.com/robots.txt to confirm the rule is live. It states a preference the big firms say they will follow from 2027.
Good to know: Blocking training does not remove you from search. It only tells AI makers not to use your pages to train their models, and private deals can still override it.
Your company data could be worth $100k–$3 million
There's a new channel of revenue that has just been unlocked for companies with over 20 people that have been around longer than a year.
Few know it, but AI labs now let you monetize your company’s operational data, and they're spending $15 billion this year on it to train the next frontier of LLMs. Your workflows, documentation, processes, and internal convos like Slack are extremely valuable and they're paying anywhere from $100k to $3 million for it per company.
Before you ask, all the data is anonymized and cleaned before it's sent to a Lab. Bezel (Y combinator backed) helps find a data partner that is the best fit for you.
FROM THE FRONTIER
The AI business is being built on your data, and getting cheaper by the month
The raw material. Every big model is trained on the open web, which means your pages, photos and posts. The new opt-out standard is the first time the largest firms have agreed on a single switch for that, which tells you how valuable the training data has become. See who signed on.
The money. Anthropic's filing shows a business going from about $1.2 billion to $16.3 billion in revenue in a year, with most of it from companies paying for the API. That is the scale the data feeds. See the filing.
The price war. At the same time, OpenAI keeps halving prices, so the cost of using these models in your own work keeps dropping. Cheaper models make small operators the fastest adopters. See the price cut.
The reality check. An opt-out that no one honors until 2027 is a promise, not a wall, and it protects only training, not the scraping that feeds live AI answers. Read the fine print before you assume you are covered. See the limits.
The reframe. Treat your content as an asset with terms, not a free buffet. Decide what you give the models, charge or trade for the rest, and use the falling prices to build with AI faster than the big companies can serve you.
MONEY FORWARD
Three ways to turn this week's AI news into income
1. AI-permissions setup for site owners. Most small businesses have no idea AI firms train on their sites. Offer a simple service: audit a client's robots.txt and set their training opt-out and crawler rules the way they want. See why this matters now.
2. Cheaper automations for clients. With Terra at half the old price, routine jobs like sorting support emails or writing product copy cost far less to run. Build the workflow once on the cheap model and charge a monthly fee to keep it going. See the pricing.
3. Explain the AI gold rush to a niche audience. Anthropic's IPO put AI business numbers in the headlines. Package plain-English explainers for a specific group, like store owners or freelancers, as a paid newsletter or short course. See the news hook.
GET THE EARNINGS GUIDE
Want the step-by-step version of ideas like these? Get The Aiwire earnings guide at theaiwire.co. These are ideas, not income promises, and results depend on your skills, market and effort.
IN THE KNOW
What people are actually watching and sharing
Claude in the drug labs. Anthropic shared early results from six drug companies using Claude, including Generate Biomedicines reporting about 10 times more receptor production in a redesigned therapy and Enveda cutting a review from over 25 hours to under three, which are the companies' own figures. See the roundup.
Agents are writing the code. GitHub's annual Octoverse report found AI agents started more than 40% of new pull requests in September, up from 17% in March, a sign of how fast coding is being automated. See the report.
Microsoft offers the door. Microsoft offered voluntary buyouts to about 60,000 eligible staff across the US, UK, Canada and India, more than a quarter of its full-time workforce, though it said AI was not driving the move. See the details.
India's sealed rooms. India proposed a framework to let companies train AI on sensitive data inside secure "sealed rooms," which the government claims could unlock $500 billion in value by 2030. See the proposal.
The quiet nudge. In an Anthropic study with 50 analysts, a hidden bias planted in an AI assistant swayed their decisions 86% of the time, yet only 15% thought the assistant had influenced them. Read the study summary.
TRENDING TOOLS
5 tools worth a look this week
1. Mistral Large 4. An open-weight, roughly trillion-parameter model that is strong at coding while running fewer active parameters. Free to download, though you need serious hardware to host it. Read the details.
2. NVIDIA Nemotron 3.5 Transcribe. A free, open-weight speech-to-text model covering about 40 languages, useful for captioning videos or turning calls into notes. You run it yourself or through a provider. See the release.
3. ElevenLabs SFX v3. Generates sound effects from a text prompt and blends several into one clip, handy for short video and podcasts. Free tier available, with paid plans for heavier use. See the tool.
4. Canva Craft. Canva's in-house image model plans the layout before drawing, so text in graphics comes out readable, and it is free to all Canva users. Good for quick social posts and thumbnails. Read the overview.
5. Empacer. A voice-first business assistant that drafts tasks like customer replies and waits for your approval before sending. On web and Android, with iOS planned. See the listing.
PROMPT STATION
Turn a repeated chore into an automation blueprint
With models this cheap, the hard part is not running the automation, it is describing it clearly. This prompt works in Claude or ChatGPT and turns a task you do by hand into a step-by-step spec you can build or hand to a developer. Pick something you do every week.
COPY AND PASTE THIS PROMPT
You are an automation designer. Here is a task I do by hand every week: [DESCRIBE THE TASK, STEP BY STEP]. The tools I already use are [LIST TOOLS, e.g. Shopify, Gmail, a spreadsheet]. Turn this into an automation blueprint. Give me: 1) the trigger that should start it, 2) each step in order with the input and output of each, 3) where an AI model does the work and what to feed it, 4) the one or two points where a human should check before anything is sent or published, and 5) the simplest way to build it with my current tools. Flag any step where you need more detail from me.
Tip: Swap [DESCRIBE THE TASK] for something concrete like sorting customer emails, writing product descriptions or posting to social. For a sharper build, paste one real example of the task's input and the output you would want, so the model matches your style.