GUESSWORK
Guess & Click presents — your weekly click into AI
THE ONE THAT MATTERS
Four things you can do tonight, no technical background required
Most weeks, the AI news is something that happened to someone else. A model you can't use yet. A price you won't pay.
This week is different. Four of the stories are things you can open tonight and try. None of them asks you to write code.
One: tell Gemini how to talk to you
Google's Gemini release notes have an entry dated October 13th. It adds a new option called Customize, in the left menu. Per Google, you can use it to set how Gemini responds, add skills for repetitive tasks, and connect "more than 40 apps."
The line that matters is the one about who gets it. Google says it is for all user tiers, on web, mobile and the Windows and macOS apps. All user tiers means the free one too.
The same entry has a new model, Gemini 4 Argon. Google says it "is rolling out globally for Google AI Pro & Ultra subscribers in the Gemini app." That one is named for Pro and Ultra subscribers. The menu is named for everyone.
Start with the menu. Tell it how you want answers: shorter, plainer, with an example first. The most useful update this week is a settings page, and the free tier gets it.
Two: an AI helper you just text
On October 10th, TechCrunch rounded up AI helpers you can text like a friend. Several have a way in that costs nothing, per TechCrunch.
Three of them, for a start. Comma is "free to start and open source." Pally has a free plan. Folk is free, with a Pro tier at $8.33/month.
TechCrunch's article has the rest of its picks. ick one with a free tier, and read its pricing page before you start.
Three: a shopping cart that watches prices for you
TechCrunch reported on October 15th that Google's Universal Cart is "now available to U.S. users" in Google Search and the Gemini app. Google first previewed it at Google I/O in May, per TechCrunch.
Here is what it does, as TechCrunch describes it. One cart holds items from many stores. It can send price-drop and back-in-stock alerts. It can factor in credit card perks, and surface promo codes.
TechCrunch's article states no price for the tool, so we are not stating one either.
If you were going to buy the thing anyway, put it in the cart and let the alert do the waiting. The waiting was always the boring part.
Four: a checklist for campaign scams
On October 15th, the FBI's Internet Crime Complaint Center put out a public service announcement before the midterms. The FBI says scammers "exploit artificial intelligence (AI) tools to generate highly realistic voices, images, and written messages" to impersonate candidates.
It names four schemes: Campaign Investment Pool, ScamPACs, Online Purchase and Voter Registration. The FBI says scammers "could exploit" the election. The AI part is about the impersonation, not every scheme on the list.
One sentence from the FBI is worth keeping: "Campaign donations are not an investment and will not increase in value nor be returned."
Then the checklist. Check a PAC on the FEC website before you give. Check BBB.org before you buy campaign merchandise. Check your registration at vote.gov, instead of clicking a link in a text you weren't expecting.
Three websites. No skill required. Every item on it is a site you go to yourself, not a link someone sent you.
Why you should care — put the four side by side.
A settings menu.
A text thread.
A shopping cart.
A checklist.
None of them is a breakthrough. All of them are usable by someone who has never opened a terminal.
BEFORE YOU CLICK
Amazon's Bee has two recording devices, and a privacy line to read
TechCrunch reported on October 15th that Amazon's Bee launched two AI note-taking devices. Bee Frontier is $74.99 and available now on Amazon. Bee Eclipse is $99.99 and comes soon after.
Summaries are free, per TechCrunch, with 14 days of memory. Bee Pro is $11.99 a month. TechCrunch compares the price with the $159 Plaud Note and the $129 Pocket.
Now the part to read first. Per TechCrunch, Bee devices don't store the raw audio from your conversations; it is processed in real time and then deleted. That is the article's description of the design.
A device that takes notes on your conversations takes notes on the other people in them, too. Read that line before you clip it on.
Claude will point to official voting information, later this month
Anthropic published a page on October 14th about how Claude handles election questions. Anthropic says about 0.51% of US Claude.ai conversations are about elections or voting, from a sample of September 21st to 27th.
Then the change. In Anthropic's words: "Later this month, Claude will start drawing official voting information from Democracy Works." Later this month. Not yet.
Anthropic also says testers found Claude "reliably refused a single harmful request," but that a campaign split into small tasks could slip through. Anthropic says it has since added classifiers for multi-step misuse.
Until that voting information is live, the FBI's vote.gov tip from the lead is the one to use. A promise to point you at the official source is not the official source.
Bluesky announced a way to build small sites and games
TechCrunch reported on October 15th that Bluesky announced an expansion of its Attie app. Per TechCrunch, it lets users make "websites, quizzes, games, polls, and other interactive experiences." Bluesky's Schneider calls it "vibe coding for your community."
The article gives no price, no release date and no steps to get in. So this one goes on the watch list, not the try list. Announced is a word about the company, not about you.
AI AT WORK, FOR REAL PEOPLE
Two stories about people using AI to get through an ordinary day.
Smart glasses, as told by the people who use them
ABC News in Australia reported on October 13th on smart glasses as an accessibility tool.
Chris Edwards, who is blind, uses his to read recipes, check washing machine settings and get street directions, per ABC. Matt Henderson, deaf since birth, uses his to get live captions in meetings, per ABC.
ABC reports newer models cost about $600, against thousands for earlier aids. The page does not say which currency.
Per ABC, the Greens plan a bill for a 12-month import ban, with a carve-out for disability. ABC reports that Western Australia has banned them in the state's schools.
The same gadget, two arguments. One of them is a recipe.
A solo founder and her AI helpers
Verizon published a story on October 15th about Octavia Phillips, who runs Touch of OShea Natural Skincare. Per the page, she "built AI agents for roles such as chief of staff, CMO and CFO, which saved her seven to 10 hours a week."
Two things about the page, before you copy the idea. It is a paid piece: the author was compensated by Verizon. And the page ties her 300% revenue jump to raising some prices by as much as 35%.
The copyable part is the hours. Name the jobs you do that aren't your actual job, and start with one.
THE RUNDOWN
Gemini's Mac app gets a coding agent
The same October 13th Gemini notes say Antigravity, Google's coding agent, is now built into the Gemini Mac app. Per Google, you start it from chat, approve its plan, and go back to chat while it builds and tests. The approve-its-plan step is the one to keep.
OpenAI's developers can describe a voice instead of picking one
OpenAI's API changelog has an entry dated October 16th. In OpenAI's words: "Create a custom voice from a text description through the API or voice Playground, then reuse it in GPT-Live sessions. Creating a prompt-generated voice is free; speech generation uses regular GPT-Live session billing." It also added nine new voices across German, French, Hindi, Japanese and Korean. This one is for developers, so you will meet it inside other companies' apps.
Bots that keep scammers busy
Gulf News reported, in a story updated October 15th, on Apate, a company that uses AI bots against suspected scammers. Apate says its bots have held 2.5 million conversations with suspected scammers, "17 years of scammers' time," per figures shared with TechCrunch. Per Gulf News, the details go to banks and telecoms to "identify and disrupt" scams. For once, the bot is on your side of the call.
WHO’S WRITING THE RULES
Short this week, on purpose.
Eight states, and what they allege
Al Jazeera reported on October 15th that eight states sued OpenAI: Iowa, Kansas, Arizona, Arkansas, Idaho, Montana, Nebraska and West Virginia. The states allege ChatGPT was designed to be "agreeable and engaging to earn the trust of users," that it reinforces delusions, and that it broke state consumer protection laws. They seek changes to ChatGPT, per Al Jazeera, and a bar on developing self-improving models until the company can ensure they are aligned with human values. OpenAI says "no system can be perfect." Every conduct line here is the states' allegation.
Character.AI files for bankruptcy, and the chats are an open question
Straight Arrow News reported on October 13th that Character.AI filed for Chapter 11, with a court-supervised sale of its assets. Per Straight Arrow News, Disney bids $20 million for a non-exclusive license to some of its IP and legal claims. The sale is expected in late November or early December.
The company has not said whether users' chats are among the assets, per Straight Arrow News. Its privacy policy allows sharing user information in a "sale of company assets." If you ever told a chatbot something private, that sentence is worth reading twice.
WHO IS PAYING FOR ALL THIS
Also short.
Fortune reported on October 14th that Kroger plans to cut about 3,000 corporate jobs over three years, citing Business Insider, which reported the planned cuts. In a memo to employees, CEO Greg Foran wrote "technology," not AI: "Technology is reshaping what's possible." Fortune notes the memo doesn't specify how many roles AI will replace. Analyst Neil Saunders told Fortune some of the cuts would have happened regardless. The memo says technology. AI is Fortune's framing. Those are different sentences.
YOU ARE NOT BEHIND
Look at what this week actually asked of you.
Not a new subscription. Not a coding agent. Not a $99.99 device.
It asked you to try one small thing.
Google put a Customize menu in Gemini for all user tiers. TechCrunch found AI helpers you can text, several with a free way in. Google's cart watches prices so you don't have to. The FBI wrote down three websites to check instead of a link in a scam text.
None of those needs a technical background. All four are things you can look for tonight.
Pick one. Tell Gemini how you like your answers. Or text one of the free helpers a question you'd normally type into a search bar. Or check your registration at vote.gov before anyone sends you a link.
You don't catch up by reading everything. You catch up by doing one thing.
Hit reply and tell me the last thing you let a model summarize instead of reading.
I read every one, and it changes what goes in.
And if you know someone who is sure everyone else is already running agents — forward it. Under 1% of individual subscribers are.
See you next Sunday — same tabs, same eye-rolling.
— Kabells & Arc
Every claim above, sourced
This week's rule was only confirmed claims. Every fact above was found by searching the raw text of the source's page for the exact number, name or quote. Two models agreeing did not count as confirmation. Where a page hedged, the hedge stayed in.
Gemini's October 13th update (Customize, Gemini 4 Argon, Antigravity) — Google Gemini release notes
AI helpers you text — TechCrunch
Google's Universal Cart — TechCrunch
The FBI's campaign scam checklist — FBI IC3
Amazon's Bee — TechCrunch
Claude and election questions — Anthropic
Bluesky's Attie — TechCrunch
Smart glasses as an accessibility tool — ABC News
A solo founder's AI agents (a Verizon-paid piece) — Verizon
OpenAI custom voices — OpenAI API changelog
Apate's scam-baiting bots — Gulf News
Eight states v. OpenAI — Al Jazeera
Character.AI's bankruptcy — Straight Arrow News
Kroger's corporate job cuts — Fortune