buaya maya || Blogger On Blogspot

October 31, 2026 | Read online

Hey, Jhon!

Welcome back to the #1 AI newsletter in the world!

Here’s what we have today:

Alien hunters hand their telescope to AI.
Silo’s creator finally built his dream app.
How the Beatles’ voices got pulled apart.
Old food papers hid a microwave pattern.

Space

AI Agents Hunt for Aliens

Source: NVIDIA

Is anyone else out there?

That’s one of those questions everyone has wondered about at some point.

In northern California, there’s a field of 42 radio dishes listening for an answer around the clock.

It’s called the Allen Telescope Array, and it’s where the SETI Institute scans the sky for signals that someone out there sent on purpose.

These days, AI agents are doing a lot of the hands-on work.

The team tells them what they need in everyday words, and the agents point the dishes where they need to go and tune the telescope until the signal comes in clean.

They also keep an eye on the data as it pours in, catching mistakes and fixing them on the spot.

That’s no small task when the telescope takes in roughly 60 HD movies’ worth of data every second.

That same setup caught a pulsar in the Crab Nebula as it happened, the first time AI has spotted one in real time.

(A pulsar is what’s left after a huge star collapses. It spins fast and sends out radio pulses like a lighthouse.)

It wasn’t aliens, of course, but it shows the agents can pick a real signal out of all that noise while it’s happening.

The SETI Institute’s head of the search even thinks the AI might end up having more in common with whatever it finds than with the people who built it.

“But this isn’t something to fear,” he said. “Rather, it’s a breakthrough in translation.”

Our take: A pulsar made a fair first test, since it’s a known target that flashes on a schedule. Once agents start picking where to point next and which blips deserve a second look, the search won’t be capped by how many hours people can spend at the controls anymore.

Together with Base44

Build Your Dream App by Describing It

Over at SETI, the team says what they need and the agents handle the dishes.

Getting an app built for something in your own life works a lot like that now.

Take a book club that wants one place to log its picks and vote on the next read. Without someone who codes, it ends up as a group chat with a pinned message.

Base44 builds your dream app from a few lines describing what you want.

It asks a few quick questions first to pin down what you mean, then builds it with the logins and the database already set up, so every member gets their own account.

When it’s done, the app is live at its own link you can send to the group, and you keep changing it by chatting with the AI.

Writing Apps

Silo’s Author Finally Built His Dream App

Source: Hugh Howey

Writers can be picky about the app they write in, and Hugh Howey might be the pickiest of them all.

He wrote the books behind Apple TV’s popular Silo, and he’s tried pretty much every writing app out there.

His favorite was an old version of Apple’s Pages, back before (in his words) Apple ruined it.

Back in 2017, he wrote a blog post describing the perfect one and gave it a name, NEO.

He was asking programming teams what it’d take to build it at the time, and told his readers it might be a decade before anything came of it, so don’t hold your breath.

A try with a dev, while he was sailing across the Pacific, didn’t get far.

The internet out there was too spotty for the two of them to work together.

This year, he finally built it himself, with Claude doing a lot of the coding alongside him.

You can tell a writer designed it.

Hit Enter once and you get a new paragraph, but hit it three times and NEO starts a new chapter, numbered and all.

Writers also get told to “kill your darlings,” the lines they love but have to cut.

NEO sends them to a tab called Darlings instead, in case you want them back later.

GitHub even called it out in their big yearly report, as an example of people who know their field building their own tools with AI.

Our take: Howey’s 2017 post reads a lot like a set of instructions. Instead of asking for “a clean writing app,” he described what happens when you use it. Move the mouse left and page thumbnails slide out, and click the word count and it switches to the current chapter’s count. That’s the level of detail AI coding tools need to build what you picture. If there’s a tool you’ve always wanted, write down how you want it first.

Coming in Hot

AI Tools of the Day

🎬 Clueso remakes any video you love in your own brand, with a fresh script, AI voiceover, motion graphics and translation in 150+ languages.
🎧 Veridive searches YouTube talks and podcasts for whatever you ask and links each finding to the exact second it’s said.
🗓️ Ariso gets to know the people you deal with, then lines up your schedule and reminders and sends you a morning brief by text or in the app.
🪧 Oginify takes a single link and designs four preview cards for it, the picture that pops up whenever someone shares it.
🤖 Survival LLM keeps an offline assistant and 1.5 million reference articles on your computer for blackouts, with a source behind every answer, free on GitHub.

Interesting AI

Candlelight Without the Candle

Source: Digital Candle

It started with a few phone videos of one candle burning in a dark room.

One of the takes was even shot in slow motion.

Those clips taught an AI how a real flame moves and where it dips and jumps.

That AI now runs inside Digital Candle.

It’s a slim board about 10 inches tall with 144 amber LEDs lighting it up.

Everything happens on the board’s own chip without any internet.

The model is tiny by AI standards and still makes the light move the way a real wick does.

Each board also builds its flicker around a factory ID burned into its chip.

That means the one on your shelf won’t flicker quite like anybody else’s.

Sound

Why the Beatles Sound Clean in Get Back

Source: MIT

Old recordings come with a funny problem.

The voices and instruments all land on the same tape, tangled together, and there’s no slider for turning one of them up later.

That’s what the team behind “The Beatles: Get Back” ran into with the band’s old studio recordings.

To clean them up, they used AI built by an MIT professor who’s spent years teaching computers to listen.

At a loud party, you can tune in to one voice and let everything else fade into the background.

His AI does the same thing with a recording, except it pulls each sound out onto its own track.

Once the voices and instruments are separated, editors can clean up one of them without messing up the others.

That work helped the show’s sound team win an Emmy.

Pointed at a busy street, the same kind of listening could pick out squealing tires and crunching metal and flag a crash.

Run it over a recording of a full football game, and it could find the highlights by listening for the moments the crowd goes crazy.

His newer work goes the other way, with AI that listens to someone play and gives them feedback on what they played.

Our take: AI that makes music gets most of the attention, but the listening side might end up mattering more to regular people. It’s the same kind of skill behind cleaner phone calls and a smart speaker that can hear you over the TV, and untangling Beatles tapes from 1969 shows how good it’s gotten.

Kitchen Science

Old Food Papers Hid a Microwave Pattern

Source: Penn State

Put a bowl of soup and a slice of pizza in the microwave for the same minute, and they won’t come out the same.

Part of that comes down to two numbers food scientists track for nearly every food.

One shows how strongly a food reacts to microwaves, and the other shows how much of that turns into heat.

Those numbers are spread out across a pile of research papers.

A food engineer at Penn State says it used to take six months or more to gather them up by hand.

His team handed that task to a group of AI agents instead.

They read through the papers and pulled the numbers out, checking their work as they went.

Half an hour later, the team had more than 4,000 measurements sitting in one place.

Once everything was lined up, a pattern showed up that nobody had reported before.

Across more than 100 foods, those two microwave numbers moved together in the same predictable way.

He figures nobody caught it sooner because no one could collect that much data by hand.

People weren’t out of the picture, either.

Food engineers chose which papers to use and checked the AI’s numbers against their own.

Our take: Pretty much every field has answers like this sitting in old papers, waiting for someone with the time to line the numbers up. If agents can do that reading for scientists, a lot more of them could turn up in places people already looked.

Prompt of the Day

Vote Off My Habits

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

View and copy today’s full prompt

This prompt runs eight of your daily habits through a season of Survivor, with the AI as the host. You finish with three habits voted off, one exit carried out before the chat closes, and a sole survivor to double down on tomorrow.

That’s all for Saturday!

Before you go, describe the app you’ve always wished existed to Base44. Give it a few lines about what the app should do and who’ll use it, and see what’s waiting for you when it’s done.