"Inspired by" might be better than "original pot" because as anyone can tell (including you!), they aren't 1:1 matches.
Almost any 3D artist doing objects, especially a new category of objects they haven't done before, will do so with 3 or more reference images from every angle and model based on those, pretty tightly initially at least. But even with that, you add your own style and taste into the modeled object, and almost never will say that what you referenced from is the "original object" in that manner.
"Inspired by" works well though and doesn't remove anyone's agency or lessen their "creation" of it :)
Melitta/Friesland made it in different sizes. The smallest is squished-down, matching the 3D model's proportions a little bit more.
I have one at a prominent place in my home, next to a Luxo Jr lamp and ball, SGI badges and autographed photo of Lenna. Still looking for a terracotta rabbit.
LIDAR data from Europe has been used to identify Celtic agricultural structures (fields) by a retired museum director. He’s also found some Roman age artifacts.
The device used to make oligonucleotides is pretty much an inkjet printer, from there it's cleverly applied enzymes to assemble the full strand.
Usually the synthesis of oligonuclotides is done with phosphoramidite chemistry [0]. So tell us more about those enzymes to make DNA molecules from scratch.
I’m pretty sure they were talking about how you can combine synthesized oligos into longer fragments when you are starting with smaller oligos. I’m guessing that they were referring to using different restriction enzymes from that point on.
Unless we are talking about really long oligos (which is harder and $$$), you’ll need some post-synthesis enzymatic work to get your full strand.
Yes I also read the comment that way. The inkjet printer remark presumably refers to benchtop DNA synthesis machines which have been commercially available for a while now. (The linked article mentions the possibility of buy-back programs for old machines that don't enforce any restrictions.)
I guess it's not a bad idea to put basic hurdles in place but it would be a serious mistake to see this as any sort of comprehensive preventative. DNA just isn't difficult to manipulate.
The vast majority of commercial synthesizers aren't inkjet. The OG DNA synthesizers are pretty simple - just a handful of pumps and valves and a packed silica column for each run. Anyone with a little mechanical engineering and organic chemistry training could build one (lots of companies in the past did just this.)
Kind of like owning your own darkroom, back in the day of analog photography.
Seeing his adjacent comment I stand corrected. I was not aware that there are now literal DNA printers. Only that you could obtain the sort of synthesizers to which you refer which had approximately the form factor of an inkjet (ie a small box).
The thing is DNA just also isn't dangerous. It's fragile and everything in the environment destroys it - as noted it's literally a everything else which is the hard part.
This reeks of the Longtermists getting their influence into government. They were doing an entire round on YouTube about a year ago.
And yes, phosphoramidites are what you use as ink. You print many many dots on a flat surface, and then you print over the same dots again and again, building up layers one base at a time. Then you dissolve those dots, giving you short strands in solution, which can be assembled into longer strands by being strategic about what else is in that solution and about which sequences you programmed in to the printer.
I've been considering making one in my garage. I don't think it's going to be easy, but I also don't think it's going to be difficult enough to be a good candidate for use as a regulatory chokepoint. We're not talking about a warehouse full of centrifuges here, it's a printer with special ink.
Ansa Bio uses an engineered terminal deoxynucleotidyl transferase (TdT) enzyme to synthesize the oligos themselves, rather than using phosphoramidite chemistry. Each enzyme is tethered to a single nucleotide, allowing it to extend the DNA strand by exactly one base. The tether is then cleaved and the cycle repeats.
This works in aqueous solution under mild conditions and enables synthesis of much longer oligos (~1,000 bases), which can subsequently be assembled into 50 kbp constructs.
Please make this more generic, using the OpenAI API standard with model discovery. That way it works against a bare Llama.cpp instance out of the box, the local inference server that many others wrap around (like Ollama and LM Studio).
This looks and works great, super impressed. This is what Siri/Apple AI should be like.
One concern I have is about sending private info to Anthropic. So it would be great to have the option to use local AI.
Local options should support the OpenAI API that seemingly is the standard for these connections, as provided by Llama.app and Unsloth Studio for example. That way one could run against a local model and keep personal information private.
That of course begs the question if a local model is smart enough for the tasks at hand.
I'm also trying out OOMU, a similar but closed source personal assistant app for MacOS. It emphasizes privacy with local models, and even has an 'Airplane' mode. The app has a built in llama.cpp server and uses Gemma-4-4B, their own router model plus a vision model.
Most daily commands like launching and moving around sites/apps happen without AI, but I agree with the privacy concern and will definitely look into utilizing local AI.
Smaller models, like Haiku, I have found to already struggle during multi-step tasks, so I expect local model capabilities to reduce performance further.
I'll look into how OOMU splits their work up, but what local model would you suggest starting with?
The choice of model could be up to the user, depending on the available RAM on their M class Mac. Newer and bigger usually being smarter.
Llama.app/.cpp and Unsloth Studio on Mac already provide guidance to pick the optimal model for the available memory, all model wheights hosted by huggingface. Both apps can use the local huggingface model cache locally.
If you make it configurable, the user could also run a separate machine with an NVidia card and Linux to serve tokens on the LAN. Or use OpenRouter or any other cloud hosted model offering (DeepSeek, Z.ai, etc.).
I would probably run qwen 27B or qwen fn but I would run the actual inference on one of my other boxes and not on actual desktop so initially it could be sufficient if you just enabled openai or anthropic compatible endpoint.
[0] https://spectrum.ieee.org/behold-the-worlds-most-famous-teap...
reply