A New Type Of LLM On The Block: Decision-Making Models

Large language models (LLMs) output language, but they are commonly tasked with making a decision or classification of some kind instead of writing an essay or chat reply. An LLM will be provided with input, and asked to classify that content in some way: with a rating, yes/no answer, a best-fit categorization, and so forth. A recent new type of model by the name of Jev was released only weeks ago and it is extremely fast, ultra-cheap, and laser-focused on that decision-making role. It can’t write even a single sentence, but it can classify and categorize very, very quickly.

Jev works like this: it still accepts text input, but it outputs only floating-point numbers. Those numbers are the “answers” to user-specified yes/no type questions, lists of choices, and scoring-type requests. [Simon Willison] provides a concise summary of what Jev does, and what makes this new category of model so interesting.

To say that the idea has caught on would be a wild understatement. Folks are making their own decision-type models and experiments in a flurry. Kev and Nimble are two examples (Nimble was added as a supported model in Ollama just recently, and is small enough to run locally with relative ease.)

If this type of local AI model was the missing link you needed to get an idea working, don’t keep it to yourself! Tell us all about it on the tips line.

A Good DIY Solder Stencil Begins With A Cleanly-Sliced Soda Can

Solder stencils are a fantastic way to accurately apply solder paste to a PCB. Professionally-made stencils are cut from steel, but for the home hacker, soda cans continue to be the alternative of choice. The only trick is how to actually get those little holes made, and [Saheen Palayi] shows off both his method of laser-engraving the holes, as well as a tool for cutting the can in the first place.

A good result starts with cutting a can cleanly, thanks to a 3D-printed tool.

To turn a soda can into a flat-ish aluminum sheet means cutting off the ends and slicing it open. Ideally, one achieves this without kinking or bending the thin metal. In practice, this is quite difficult. [Saheen]’s solution is something a little like a pipe cutter — a 3D printed tool that gradually presses a blade section from a utility knife into the empty can as it turns until it cuts through. Once the top and bottom are off, it’s easy enough to snip down the side to get a curved sheet.

The hole pattern comes from one’s PCB design software of choice, and the actual cutting is done by a fiber laser. The wavelength of fiber lasers makes them good at marking and cutting metal, and [Saheen]’s laser takes almost no time at all to cut the stencil into the thin metal.

Fiber lasers used to be the sort of tool that only industrial shops had, but they’re a prosumer-level tool that can be bought online nowadays. [Saheen] uses an xTool F1 Ultra to cut his solder stencil, and we’ve previously seen that same laser used to create a PCB by blasting away unwanted copper until only the traces remain.

We’ve seen aluminum cans etched and also machined to create stencils, and the laser is certainly the fastest. Watch it in action in the video below.

Continue reading “A Good DIY Solder Stencil Begins With A Cleanly-Sliced Soda Can” →

Fixing A Power Grid That Loses Over Half Its Power To Inefficiency And Theft

[Mini Shaji Thomas] has a deeply interesting article on IEEE Spectrum detailing how the power distribution system for a city of over twenty million people went from failing to first-class. In the early 2000s, over half of Delhi’s generated power was lost to inefficiencies of all kinds, and theft. Losses were technical as well as administrative, and the rampant theft was perpetuated by “both the powerful and the powerless”. Spoiler alert: it’s a big job.

That level of loss is staggering, and caused by many issues rather than just one or two. Many of the problems tended to compound one another. For example, an effective power grid relies on balancing active and reactive power, but if that is not well managed then inefficiencies cause losses and outages that tend to lead to even more losses and outages as the system is never able to catch up to demand. On the scale of tens of millions of inhabitants, such losses are far from trivial.

There wasn’t a silver bullet solution. Modernizing and stabilizing the power distribution required technical resources, modernized equipment, political will, expert planning, a willingness to follow through, and authority.

The starting point was grim: a failing network that was constantly overloaded, poor or nonexistent controls and safeguards, inaccurate records and inefficient administration, no real means of sensing faults or inefficiencies or theft, weak to nonexistent enforcement, and an untrained workforce. Still, it could be done, and it was. While there’s still improvement to be made, revamping the system was ultimately successful.

Individually, some of the solutions were quite simple. For example, replacing bare overhead distribution wires with a system of insulated cables containing multiple conductors was not just safer and more resilient, it made it much harder for people to tap into the lines. Other initiatives like making it easier and more convenient to pay accurate bills helped, too. Turns out it’s easier to have legit customers when the product and service is reliable.

Give it a read if you have a few minutes, because besides providing a peek into the technical end of how effective power grids work, it’s a good analysis of what can be accomplished with some co-operation and a willingness to change.

Sheety Turns Spreadsheets Into Binaries, For Some Reason

Ever wanted a spreadsheet as an executable with a built-in terminal interface? No? Well, that’s a pity because [Roberto Alsina]’s Sheety project does exactly that.

Sheety compiles spreadsheet definitions into a self-contained, executable binary with zero runtime dependencies. It presents a simple terminal application (with mouse support) in which one can browse and edit data and formulas with a simple interface. As far as formats go it can import and export Excel files, or a human-readable .yaml file.

When Sheety runs, the data and formulas defined are compiled into a self-contained executable with the formulas baked in, and that’s actually what one sees and uses. If a formula gets changed in the UI, a new and updated version is created on the fly to reflect the changes.

Sheety is perfectly functional, though it does have some limitations. Can the display width of cells be resized? They cannot. Does anyone actually need this? Probably not, as [Roberto] happily admits. Is it a fun project? Absolutely.

Be warned that while Sheety supports the Excel format, it understandably doesn’t support every single function Excel has added in the over forty years it has been around.

Sheety is written in Crystal and the GitHub repository has everything you need if you’d like to give it a try for yourself.

Ways To Empirically Identify A Magnet’s Polarity

Every magnet has a north and a south pole, but which is which? Sometimes it matters. If a product one builds features a magnetic closure or other part, the polarity of those magnets should be consistent in assembly. So how does one ensure they never glue a magnet wrong again? [Clough42] shows several ways to identify a magnet’s north and south poles using things many of us probably have ready at hand, and goes into a bit of theory while he’s at it.

Probably the easiest way is to use a known-good and clearly labeled reference magnet. Same poles repel, and opposites attract. But if that’s not available, a simple magnetic compass can help. Because opposite poles attract, a compass’s north point will be attracted toward a magnet’s south pole, and vice versa.

A Hall effect sensor, or an electromagnet — the winding and current flow determine the polarity — are other ways to measure a magnet’s poles. And here’s where [Clough42] dives into some details of how magnetic fields actually act, because it explains some seemingly strange behavior.

For example, at around 4:08 he demonstrates a Hall effect sensor board that is documented as lighting an LED when the south pole of a magnet is held to its front. It does that, but it also lights the LED when the north end of the magnet is held to the sensor’s back. That’s because the sensor isn’t actually directly sensing the magnet’s pole, it’s sensing the orientation of a magnetic field. The lesson is clear: make sure you’re measuring what you think you’re measuring. Near the end of the video he demonstrates a similar experience with a handy mobile phone app that senses magnetic fields by reading the device’s internal magnetic compass; by waving a strong magnet around, the detected polarity flips back and forth even though the magnet’s orientation isn’t changed.

So what does one do after positively identifying a magnet’s north and south poles? Label it clearly for use as a known-good reference magnet in the future is our suggestion. Watch the whole video below, then take a few minutes to dive into the nitty-gritty of what magnets actually are and how they work.

Continue reading “Ways To Empirically Identify A Magnet’s Polarity” →

Enormous Fluid Simulation On Flip Dots Is Also Enormous Amount Of Work

Flip dot displays are cool, and more people realize that after [mitxela]’s fluid simulation on flip dots installation was on display at EMF 2026. As glorious as the result is, it was also an amazing amount of work!

Not only did [mitxela] need to source a large number of flip dots, he also needed to find a solution for driving them that didn’t end up more trouble than it was worth. Just about everything about the surplus flip dots — from electrical requirements to mounting — was a pain to work with in one way or another. Even his optimized method of integrating a custom backpack-style driver board into the existing PCB involved a staggering amount of soldering. This project was a long time coming, and the work never really let up.

The payoff, however, is exquisite. Check it out in the video (embedded below) which really shows it off. Flip dots are like nothing else, and the subtle rippling of sound that accompanies their physical movement is oddly soothing.

The installation at EMF 2026 had a GRAVITY CONTROL joystick that allowed folks to interactively shift the display, but [mitxela] also has an accelerometer mounted so that the display physically reacts to being moved. It’s a fantastic spectacle, even more impressive in light of the work it involved.

Unsure how, exactly, flip dots work? We’ve covered all the details about how these devices function. And while a large number would be prohibitively expensive for most projects, if your project can get away with only one dot you’re probably in luck.

Continue reading “Enormous Fluid Simulation On Flip Dots Is Also Enormous Amount Of Work” →

Easy Ways To Sink A Hardware Startup

[Ryan Walker] may have written up his observations a few years ago, but the lessons are just as relevant today as they were back then. He shares five easy ways to sink your hardware startup.

It’s a reminder that while hardware startups are unique, there are basic business realities that still apply because the hardware itself is going to be only one part of a whole. These business fundamentals can be a drag, but it’s worth giving them some attention. But if that’s not your jam, no worries. As [Ryan] experienced, they will explain themselves one way or another.

A good one is skipping market research. Do customers actually exist for this thing? Or forgoing market testing — do the customers actually want it enough to pay for it? One of the worst things to be stuck with is a product that everyone likes, but nobody wants to buy.

Premature optimization is another good one that a number of our readers can probably relate to in one way or another. It’s one thing to buy a tool or a part that one doesn’t end up needing, but when that gets scaled up it can put a real dent in a fledgling business’s development.

We’ve also shared insights on what it takes to develop a product and get it out there, whether as a solo entrepreneur or as part of a larger team, to help nudge the process toward success.