AI at Work

You can now build a good-looking website by borrowing a designer's rulebook

Building a decent-looking site used to need design skills or money for a designer. Now you can grab a ready-made design guide from a free library, hand it to an AI helper on your computer, and have it build pages that follow that professional style. For a student or job-seeker, that means a real portfolio site this weekend instead of someday.

For youBrowse the free Refero Design library, pick a style you like, and have an AI build a simple personal site that follows it. https://styles.refero.design/

Source: Refero Design ↗

AI at Work AI Productivity Tools

As AI records every meeting, protect what only humans do there

Soon every meeting will be recorded and written up for you, so taking notes is no longer the point. What is left for people is the part a transcript cannot do: setting a clear agenda, reading the room, drawing out the quiet voices, and naming the decision before everyone leaves. One founder who studied top companies runs fewer and shorter meetings, and guards a no-meeting day so people can actually do the work.

For youPick one recurring meeting this week, add a three-line agenda and an end-early rule, and let an AI note-taker handle the write-up so you can focus on the discussion.

Source: Charter ↗

Hand the AI your spreadsheet, get back a first draft of your slide deck

Turning a messy spreadsheet into a presentation used to eat an afternoon. Now you can upload the file, tell the AI what question the deck should answer, and get a full draft with charts and speaker notes in about ten minutes. The work that stays yours is the judgment: which story the numbers really tell, and which recommendation you are willing to stand behind.

For youTake a spreadsheet you already have, open Claude Design, ask it to build a short deck on what happened and what to do next, then rewrite the one slide that carries your real recommendation.

Source: Anthropic ↗

AI at Work AI Governance & Regulation

Your website now has to decide which AIs get in

Until now, any AI tool could quietly read your website, whether to help someone search, to act on their behalf, or to train itself on your words, and you had little way to tell those uses apart. Cloudflare just split that access into three separate switches: search, acting for a person, and training, so site owners can allow one without allowing the others. Starting in September, new sites on Cloudflare will block the acting and training kinds by default, while still letting search bots in.

For youIf you run a website, blog, or online shop, spend ten minutes this week checking your site's AI traffic settings and choose on purpose, rather than by default, which of the three you want in.

Source: Cloudflare ↗

AI at Work Organizational Change Skills & Reskilling

The AI skill companies are now paying for is making it actually work

For the last two years, AI vendors mostly sold you access to their AI and left you to figure out the rest. Microsoft just put $2.5 billion behind a different bet: a new 6,000-person team whose whole job is sitting inside client companies and building the specific AI tools those companies actually need, following Amazon's similar billion-dollar move days earlier. At Cisco, that shift already shows up in daily work: its finance team now gets 80 to 90 percent of first-draft SEC filings written by AI, with people reviewing and signing off on the rest.

For youPick one repetitive task on your team this week and write out, step by step, exactly how you do it. That map is what makes you the person who can point AI at it well.

Source: GeekWire ↗

Most people use AI at work now. Most of the payoff is going to babysitting it

87 percent of office workers now use AI at work, and most say it saves them real time, roughly 11 hours a week. But Glean's Work AI Index 2026 found only 13 percent of companies see a real improvement in results, because workers spend an average of 6.4 hours a week checking AI's answers, fixing its mistakes, and feeding it missing information, a hidden job the report calls botsitting. The time you save by using AI is only real once you count the time you spend watching over it.

For youFor one week, keep a simple tally of the minutes you spend double-checking or fixing anything AI gives you. That number tells you whether AI is actually saving you time, or just moving the work somewhere less visible.

Most people feel the boost personally. Almost none of it shows up as real results for the business, because 6.4 hours a week go to checking AI's work.

Data: Glean, Work AI Index 2026

Source: Glean ↗

Source: McKinsey & Company ↗

Companies bought the AI. Most still haven't redesigned the work.

More than 80 percent of companies say their spending on AI has not yet shown up in better results, according to McKinsey. The problem usually isn't the technology, it's that most workplaces added AI on top of the old way of working instead of redesigning who does what and when. That gap, between owning the tools and actually changing the job, is where the real disruption and the real opportunity both sit.

For youAt your own job, stop asking 'which AI tool should I use' and start asking 'what step in my process could disappear completely if I rebuilt it around AI from scratch'.

AI at Work Demographics & Labor Robotics

Being good at AI won't save your company. Rebuilding around it will

Most executives have treated AI like a new piece of software to install, something the IT team rolls out and everyone gets trained on. McKinsey's leadership research argues the companies that win are treating it as a full rebuild of how the business runs, since as much as 80 percent of what makes AI pay off is redesigning the work itself, not the technology behind it. That means a CEO's real job in this shift is deciding which parts of the company change first, and having the nerve to actually change them.

For youAsk your own leadership, or yourself if you run the show, which single process would change the most if you rebuilt it around AI from scratch, instead of bolting AI onto how it already works.

Source: McKinsey ↗

China's rented robots still need a human at the controls

You've probably heard that Chinese factories already run on armies of tireless robots. A new investigation found the reality is messier: even the most advanced humanoid robots there still need a person watching over them, and can only match about 80 percent of a human's output on simple, repeated tasks. The robot rental business renting these machines out by the day is booming, but it still runs on human operators as much as on the machines themselves.

For youIf robots start showing up in your industry, the safer bet is learning to supervise, troubleshoot, and work alongside them, not assuming they'll simply replace you.

Source: CNN ↗

Stop opening a new tab, just tag the AI where you already work

Until now, using AI at work usually meant opening a separate app, asking your question, then copying the answer back into Slack or email. Anthropic's Claude can now be tagged right inside a Slack channel like a teammate, given a task, and it works in the background and reports back when done. Setting it up takes an admin a few steps, connecting the tools it can use and choosing which channels it can see, but once it is running anyone on the team can hand it work with an @ mention.

For youPick one recurring task you currently do by switching to a separate AI tab, like drafting a weekly summary, and check whether your team's tools now let you request it from inside the app you already use.

Source: The Rundown ↗

Source: Thinking Machines Lab ↗

Why a tiny AI beat the biggest models by learning your judgment

Bridgewater tested the biggest AI models on the kind of judgment calls its analysts make every day, like deciding which headlines matter. GPT, Claude, and Gemini variants only got it right about half the time, even with the fund's own experts writing careful instructions. A much smaller AI, trained directly on examples of the experts' own decisions, scored 84.7 percent right, at 13.8 times less cost, because it learned the judgment instead of just following instructions.

For youStart writing down the judgment calls in your job that a general chatbot keeps getting wrong, since that record is what turns your expertise into something an AI can actually be trained on.

Training a small AI on your own expertise beat prompting the biggest models, at 13.8x lower cost.

Data: Thinking Machines Lab & Bridgewater

Checking the AI's work is becoming the real job

Amplify's 2026 AI Engineer Survey found that 95 percent of AI engineers now use AI agents at work, nearly double last year, and 89 percent let those agents write or change real data, not just draft text. The tools for controlling what agents are allowed to do are still basic, mostly a human clicking approve, and 59 percent of teams worry the AI-written work is quietly piling up as future problems. Even inside Anthropic, one executive said his own team is now bottlenecked on review, on finding the time to actually check what the AI already did.

For youBefore you let an AI agent take an action on your behalf, like sending an email or updating a record, build in a quick review step, because checking the work is quickly becoming the real job.

Source: Latent Space ↗

The free-spending era of AI at work is ending

For the past couple of years, many companies let employees use AI tools with almost no limit on cost. That is changing fast: Walmart, Uber, and Microsoft have all started capping how much AI usage employees get, after Uber reportedly burned through its entire annual AI budget in just a few months. Procurement teams are now asking every team to justify what each AI tool is actually worth, not just how often people use it.

For youStart keeping a simple note of what your AI tool use has actually saved you or produced this month, in time or results, so you have a real answer ready when someone asks whether it's worth the cost.

Source: MarketScale ↗

Teach your AI where you keep your life, then hand it real tasks

One person needed a taxi booked while traveling, so he had his coding AI check his calendar for the flight, search his email for the address he used last time, then go online and complete the booking and payment itself. The trick was not a clever instruction. It was that his files already held his memories, his ongoing projects, and notes about himself, so the AI had everything it needed without him explaining from scratch.

For youStart one plain text file today that lists your recurring tasks, key facts about you, and where things live, so any AI tool you use later already knows the basics.

Source: Ben's Bites ↗

Reviewing a tax return beats typing one in by hand, and that is now the job

Accountants using OpenAI's new tax tool no longer start from a blank form. The AI reads the messy PDFs, spreadsheets, and notes, prepares a full draft, and shows exactly which document and cell each number came from. The team even found cases where the AI's number was right and the old human-entered answer was wrong, so the job shifted from data entry to checking evidence and catching what the AI got wrong.

For youNext time an AI hands you a finished draft, ask it to show its sources before you approve anything, the way you'd want to see a source anyway.

Source: OpenAI ↗

Source: The Information ↗

AI help is moving into the chat apps you already use

You used to open a separate AI website, ask your question, then copy the answer back into Slack or Teams. Anthropic is now building Claude into Microsoft Teams too, so people can tag it right inside a group chat and get help without switching tools. Microsoft and Salesforce are both letting a rival's AI operate inside their own apps, because keeping people in the app matters more to them than keeping a competitor out.

For youNext time you need help on a work chat, check if you can now tag the AI right there instead of opening a new tab.

Claude's everyday AI can now finish multi-step tasks on its own

Until now, an AI that could work through several steps on its own, browsing a page, testing a fix, filling out a form, lived behind the most expensive plan. Anthropic just brought that same follow-through to Sonnet 5, the AI most people get by default when they open Claude for free or on the cheaper Pro plan. Early testers had it investigate a bug, write a test, fix the code, and confirm the fix worked, without checking in at every step.

For youNext time you use Claude, give it one task with several steps instead of breaking it into separate messages, and see how far it gets before you need to step in.

Source: Anthropic ↗

Stop organizing your AI with files and folders, organize it with two simple tools

People setting up Claude for real work used to build a folder full of instruction files and hope the AI read the right ones at the right time. That approach breaks down once more than one person needs to use it, so the better setup now is two things: a Skill for anything you do repeatedly, and a Project for anything tied to one client or context. Teams that keep the two separate can share and reuse both without the AI dragging outdated information into the wrong conversation.

For youPick one task you re-explain to an AI every week and turn it into a saved Skill this week, instead of writing it out again from scratch.

Source: Ruben Hassid ↗