Hey, AI Enthusiast!

Welcome back to the #1 AI newsletter on the planet!

Here’s a glimpse into what we have today:

De-Extinction

The Woolly Mammoth Is Coming Back

Extinct means gone. Forever. No second chances.

The woolly mammoth vanished more than 4,000 years ago.

One team says forever is up for debate. Its mission statement ends on a promise.

“Next, we will have the woolly mammoth. Alive again.”

The starting point is an elephant.

An Asian elephant shares 99.6% of its DNA with a mammoth.

The last 0.4% is the whole job. Finding it is where AI comes in.

Mammoth DNA comes out of old bones in tiny, broken pieces.

Computers stitch those pieces against an elephant’s genetic code and flag every spot where the two differ.

AI and faster computing cut the work to hours.

Then comes the harder question.

Which differences made a mammoth a mammoth?

The team ran the analysis across 121 mammoth and elephant genomes to find the genes behind the shaggy coat. When they edited those genes into mice, out came 38 woolly mice with the traits the analysis predicted.

The list keeps growing.

The first plan called for editing around 60 genes. Now it’s roughly 150, including ones that shrink the ears and shorten the tail to hold in heat.

The extra genes pushed the first calf from 2028 to the early 2030s. It’ll be a cold-hardy elephant with mammoth traits, not an exact copy.

Our take: For all of history, extinction ran one way. Now AI reads what the mammoth left behind, gene by gene, and hands scientists a way back.

Together with Attio

Your Deals Go Extinct Quietly

The call went great. They said send the proposal.

You sent it. Then nothing.

A week goes by. You tell yourself you’ll follow up tomorrow. Tomorrow turns into three weeks, and the deal is gone.

Here’s the part nobody tracks. Your deals rarely die with a no. They go quiet, and nobody notices in time.

Attio notices.

Its agents chase buying signals, spot stalls before your next call, and move deals forward around the clock.

The quiet deal gets caught while there’s still time to bring it back.

Privacy

Your Anonymous Account Isn’t Anonymous

You’ve got an account under a made-up name and a blank photo.

You post there about your job, your city, the dog you walk every morning.

None of it gives you away. At least, not on its own.

Researchers wanted to know what happens when AI reads all of it at once.

They took 338 people on a popular tech forum who’d linked their LinkedIn in their bio. Then they scrubbed out names, links, and handles.

An AI agent read what was left, searched the web, and went looking for the real person.

It named 226 of them. 67%.

When it made a guess, it was right about nine times out of ten. When it wasn’t sure, it stayed quiet.

It didn’t hack anything. Every clue came from posts people wrote themselves.

A second test on movie fans showed the pattern.

People who mentioned one film were hard to match. People who mentioned ten or more got matched almost half the time.

Our take: Your posts are a puzzle, and every detail is a piece. The risk is a stranger piecing together who you are. Before you post, ask whether a detail points straight to you, like your street, your kid’s school, or your daily routine. Fewer pieces, harder puzzle.

The catch: The test picked people who had already linked their real profiles once, which made them easier to find than most anonymous users.

Wispr Flow In Focus

Say It Rough. Send It Right.

You write the message. You read it back. It sounds wrong.

So you reword it. Now it’s stiff.

You try again, and ten minutes later you’re still stuck on the same three sentences.

Here’s the thing. You already said the right thing. It came out in the wrong shape.

Wispr Flow’s Transforms fix the shape.

Say it out loud or type it, highlight it, and hit a Transform.

Polish tightens it for clarity. Prompt Engineer rebuilds a rough ask with the role, task, key points, and tone.

Same idea, sounding however you want.

Science

An AI Built Its Own Lab

Ask AI the same question twice and you’ll often get two different answers.

You pick the one you like and move on.

An MIT professor did the opposite.

Markus Buehler handed an AI agent five pictures of natural materials built in layers, plus one question.

How does all that layering change the way a material bends, cracks, and breaks?

The agent skipped the guessing. It built a lab to find out.

It wrote a shape generator, a physics simulator, and a crack tester, then wrapped them into an app you open in a browser.

It planned its own experiments and wrote down its predictions before running them.

Then Buehler ran the whole thing again. And again.

Three runs. Three different labs. Partly different conclusions.

One finding showed up every time.

More layers didn’t reliably make a material tougher.

What mattered was how the thick and thin pieces were arranged and how they shared the load.

Later, hundreds of AI agents moved into the labs and kept experimenting.

Our take: One answer is a guess. The same answer three times is a pattern. Next time the stakes are high, ask your AI the same question in three fresh chats. Trust what repeats, and dig into what doesn’t.

The catch: Every result came from simulations the AI built itself. Nothing was made or tested in the real world yet, and the paper isn’t published.

Coming in Hot

AI Tools of the Day

Workplace

Museums Are Winging It With AI

You paste a client email into a chatbot to clean it up.

Nobody told you if you’re allowed to. Nobody showed you how to check what comes back.

Museums are doing the same thing.

A global survey of more than 400 museums in 90 countries found 57% already use AI.

They lean on it for admin work (70%), communication (62%), and research and record-keeping (43%). Some use it to build exhibits and engage visitors.

Now the gap.

55% have no AI policy, strategy, or guidelines. Only 17% train their staff.

Most of the use is staff-led. People picked up tools on their own and figured it out as they went.

Their top worries line up with yours. Wrong facts, copyright, data protection, and bias.

When asked what they need, museums named practical guidance, training, and a way to trade what works with each other.

Our take: The experimenting is the healthy part. The missing piece is a shared playbook. If your job has no AI rules yet, draft a one-page version with your manager. What’s safe to paste, what gets a second check, which tools are approved. Clear rules make people faster, not slower.

Notable AIs

Notable AI Tools

🗂️ MongoDB MCP Server hands your coding assistant your live database, letting it build against today’s data instead of field names from last week.

Interesting AI

So, you meant to log lunch again. But another call ran long, you grabbed a handful of almonds, and by dinner the day became a blur.

Food apps want a photo of every bite, and almost nobody keeps that up.

ODYSS N1 does the logging for you. It’s a necklace with a small wide-angle camera.

It notices when you eat, figures out what’s on your plate and how much, and scores the meal. Before dinner, it suggests a simple next step, like more fiber.

A physical button shuts off the camera and mic, and photos get deleted once they’re read.

Policy

AI Has a New Name in Washington

You’ve called it AI for years. The White House wants a new word.

During a three-day state visit last week, the US and China agreed to call the tech “super intelligence,” or SI.

President Trump first used the term in a speech at the UN earlier the same week. China’s foreign ministry said it valued the U.S. using the new term.

The two sides also set up a U.S.-China Super Intelligence Dialogue to talk through the technology’s risks and benefits, with the first exchange due by November.

They added a direct channel for SI incidents too.

Neither side announced a joint plan for building or regulating the most advanced models.

Our take: For you, nothing changes. Your chatbot, your apps, and your prompts work the same tomorrow. The November talks will show whether the new word comes with new rules.

Prompt of the Day

Build It or Find It

Important Note: You’ll need to click the button to get the complete prompt.

That’s all for this Sunday!

Thanks for reading,

— There’s An AI For That