If you’ve been building features with LLMs over the last few years you’ve likely used frameworks like LangChain/LangGraph, Vercel AI SDK, LlamaIndex, Google ADK, etc. Alternatively, you may be deeply exposed to chatbots, coding agents, and/or assistants. All of the above, roughly circling around the same idea, are flavors of the AI Agent loop being a mix of large language models and software scaffolding (typically called a harness).

What’s surprising is that for Swift and Apple platform developers, a solid version of the harness-building toolset doesn’t seem to exist yet. Working on some exploratory features, I kept having to relearn and rebuild the same scaffolding every time I wanted to try a different provider. The kind of thing a framework exists to solve. And since a clean Swift abstraction of it didn’t already exist, I built mine.

While I flattened the repository history before making it public, the repository history of AgentKitten spanned more than 570 commits and 130 pull requests of rethinking the shape of things. And I've continued to maintain it since, having piled up 64 more PRs to improve it further. It is built using AI coding tools, but steered carefully and deliberately.

I believe we still have a world of opportunities to put AI (particularly small on device models) to the service of improving mundane interactions in our apps (I put some samples in the repo’s README.md). I think this focus was kind of instrumental on how I built AgentKitten. It is more powerful that I expected it to be when I started it, but still I tried to stay grounded about being practical for the little use-case.

Clone it and run `swift run Playground --help` to see examples of what it can do.

Without widespread adoption is still an early stage project and I'm looking for constructive feedback to improve it.

My thoughts on the use of AI and its relationship with human reviews. A little toy sample sprinkled on the post to try illustrate the point.

For those non familiar to Ook! http://www.dangermouse.net/esoteric/ook.html

That somehow reinforces that publishing source code with CS research papers is more than necessary :)

Indeed, it makes me wonder if published code leads to higher citation counts.

Although if it does, I suspect it has more to do with the fact that there's a product & marketing around it.

I wonder if it's more subtle than that - if your paper has associated source code, then it's likely that people reading it might try it out, and the ideas that you've presented will stick around more than a short paper with no follow up material.

In other words, papers are only a short glimpse into your research, presenting code allows an interested reader to look deeper, and means that they're more likely to remember what you've done, and cite it later.

I don't think that they even expect them to be predictors of importance or popularity. The "best paper" award can give a hint of temporary fashions or about what seemed to be interesting concepts/works at a certain point in time. That's why it seems especially interesting to see them compared with the "top cited" in these charts.

I'm nowhere near to be techno-apocalyptic, but I just thought that from a "1984" point of view it could be a tool to monitor activities without no permission at all.

And wearing a tin foil hat just makes you a better reflector!

Anyway, if you look up through-wall radar detection, most of the implementations use S-band radar, and it is probably safe to assume that it's already been figured out how to use wifi stations and devices already in place emitting signals to each other, rather than always needing to bring additional equipment.

With regard to privacy though, it seems like it would be data and computationally intensive to track the movements of people in a large number of houses.

Let's say that a hypothetical application could only detect velocities away from or toward the base station, but no directional information. With each movement you make it would build a set of hypotheses about your location. Then when you make another movement, it would need to branch out, and create an entire set of hypotheses for each hypothesis in the previous set, and so on with each movement. While the program could abandon entire branches when they showed you walking through a wall, it would still grow very quickly.

Though it is possible to mine data of cell phone locations and CC television cameras after the fact, it seems like (if my understanding of how the technology would work is anything near reality) it is simply too difficult to collect a lot of information about people's movements in their homes without a specific prior reason to justify the expense.

Anyway, perhaps rather than tinfoil hats, people worried about surveillance should hang lots of mobile sculptures in their houses to create confounding noise instead.

Not supported on iPhone 3GS: no front-facing camera. I didn't expect that given "ALL your friends can use Hangouts" and "and no one gets left out".

Absolutely necessesary to share it along with http://runningastartup.tumblr.com/ :P

Nice! I'll definitely try it even though I don't have two Macs :P

Don't you think that specific industries may need specific tools that Kickstarter doesn't provide?

And the "return" is absolutely different...

I think that these simple things are the sparkle to create the need for separated platforms.

However, many of these alternative croudfunding platforms are not implementing any difference. And there I will support your point of view, there is no need to do that. In fact, it should be better to take advantage of Kickstarter's moment of inertia.

I can't really think of a good example of a tool that is needed to crowdfund a project in a specific industry.. but this is the best argument I've heard and I can imagine something like this actually actually adding some value.

There are many things I'd expect as such tools/customisations... A pair of simple examples:

* Specialised support: the croudfunding platform could offer legal and strategic support in the narrow field they are focusing. I think Kickstarter cannot embrace all industries even if they try.

* Specialised tools and rules: Each platform must have extremely distinct interactions with the backers and product outcomes. In an academic platform (as the one being discussed) I'd expect some tools for scientific divulgation (which must be a mandatory "return" from crowdsourced research projects) and collaboration (which must be extraordinarily stimulated). I think Kickstarter will never consider/establish these rules/tools.