• Apple's Home Run
  • The (AI Task) Force Is Not With Us
  • Vibe Coding Goes Mainstream

Apple's Home Run
Apple has had a few toes dipped into the smart home sea, but this week they jumped in with both feet. Expanding their line from the HomePod, Apple announced the new Amazon Echo and Google Nest Hub competitor, the $300 HomeView. The HomeView looks a lot like a 7" Apple Watch sitting on a base that is intended to be reminiscent of the iMac G4 from the early 2000's. As with that iMac, the screen on the HomeView can be tilted up or down for easier viewing. The screen is more square than rectangular, which is an interesting idea for a product that lives in a world of very rectangular screens optimized for watching video. The HomeView does not really seem intended to watch video; the device does not have an app store and the pre-installed apps include FaceTime, Messages, Podcasts, Safari, and a variety of typical functions like clocks, reminders and weather. No YouTube app is provided nor is any other video streaming app. In theory, it may be possible to access some of these tools via the browser, but it seems clear that Apple intends this device to be focused on music and podcasts, not video. The device does in fact run apps, but the apps are based on the CarPlay standard and not iOS, iPadOS, or even WatchOS. The tabletop model has two custom full-range drivers, a downward-firing woofer, and dual passive radiators. It supports spatial audio with Dolby Atmos. The wall mountable version has only the full-range drivers and is aimed more at people who want the HomeView as a central hub for their smart home. The HomeView supports Thread and Matter and Apple Home/HomeKit. The primary challenge for HomeKit has been that few of the most popular smart home devices support it. The most popular video doorbell comes from Amazon's Ring, which is not HomeKit compatible. Likewise, most of Google's smart home products also do not integrate with HomeKit. To combat this, Apple partnered with LG and Schneider Electric to create a line of products that work with HomeKit so that the HomeView can be a viable smart home hub for anyone willing to buy into the new line, replace their existing products with Apple's sanctioned products, or set up integrations with third-party options like Starling Home Hub or Homebridge.

Initial reactions to the HomeView launch have been largely tepid, with great skepticism of the pricing and features of the HomeView and the new smart home products coming from a brand (i.e., LG) that has recently struggled with privacy and security issues with their own products. The general thinking has been that anyone who is interested in smart home products has likely already bought into Ring, Google, or a mishmash of other existing products and rebuilding a smart home is not something most normal people are likely to do. This seems to suggest that anyone willing to dish out $300 for the HomeView will probably use it mainly for playing music, chatting with Siri, and holding FaceTime calls in their kitchens or living rooms. But for anyone dedicated to the Apple ecosystem, this may honestly be enough functionality for some to augment or replace Nest Hubs or Amazon Echoes with a HomeView, especially if those devices aren't serving as the central hub for their smart home devices. For anyone using a smart speaker primarily to answer questions, start timers, and play audio, the HomeView is a viable option as it links into your iCloud account so you can access your messages and have a Siri experience that is personalized to you. That puts it ahead of Amazon and Google from a personal context perspective and that might be enough to hook iPhone users.

When I first heard about HomeView, I was intrigued by the facial recognition. I live in Illinois, a state notorious for its legal restrictions on facial recognition. We have Ring cameras around our house, but they can't identify known people as they walk around our yard. The cameras can indicate that cars or people are approaching, but no further details are given. My parents, who do not live in Illinois, also have Ring cameras around their house and the notifications from those cameras can identify people by name and cars by color. We don't get this level of detail from security cameras in Illinois because under the Biometric Information Privacy Act (BIPA), facial recognition requires consumer consent, but opting into facial recognition for the HomeView is part of the onboarding process when the device is set up and paired with a user's phone. Apple also says that its facial recognition is handled on device, so it should not violate BIPA, which specifies that any biometric data stored on remote servers also requires specific consent.

Despite its price, I believe that the HomeView will be popular with many iPhone users due to its ability to connect into messaging and support FaceTime calls. The smart home features will just be a nice addition for most.

The (AI Task) Force Is Not With Us
A group of AI leaders met with officials from the United States government this week to establish some guidelines and guardrails for AI deployment. The government offices involved include the Office of Personnel Management (OPM), the Federal Trade Commission (FTC), the Director of National Intelligence (DNI, who serves as chair of the committee), and the Department of Defense (DoD). Other government officials were also invited, including Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, White House Chief of Staff Susie Wiles, Office of Science and Technology Policy Michael Kratsios, and National Cyber Director Sean Cairncross. AI companies include OpenAI, Anthropic, Google, Meta, and SpaceXAI. The specific goals of the meeting were to write a charter that lays out the goals for the task force and determine what needs to be accomplished in the 120-day evaluation window. Initial reporting from the meeting indicates that the group established their three focus areas and expanded their membership to include NSA director Joshua Rudd and Social Security Administration commissioner Frank Bisignano.

The three areas of focus for the group are as follows:

  1. Frontier lab and interagency coordination, to be led by DoD technology official Emil Michael, who will be the primary liaison between the frontier AI labs and the various government agencies. This group will be responsible for establishing rules and guidelines for reporting on AI and cybersecurity breaches.
  2. Critical infrastructure and workforce, to be led by OPM director Scott Kupor, who will be focused on securing critical infrastructure and partnering with the Department of Labor to keep watch over labor and economic impacts.
  3. Law enforcement and consumer protection, to be led by FTC chair Andrew Ferguson and the DOJ, will be responsible for monitoring fraud and anticompetitive and anti-consumer behavior.

Over the next 120 days, the force will be working on a report that focuses on cybersecurity and infrastructure protection, model control, and biosecurity. What wasn't addressed was how the group will monitor and enforce the tenets of the agreement the frontier AI companies made on September 29 concerning implementation of internal controls and external evaluation of security practices. A working group has been created to handle implementation of that agreement and the group is led by Jay Clayton, the Director of National Intelligence and the National Cyber Director Sean Cairncross. Implementing the agreement does not necessarily mean enforcing it, but time will tell whether or not the sticks used to poke the companies will be in pursuit of ensuring public safety or simply ensuring the government gets what it wants. So far, the only punishment of any kind toward any AI company has been imposing export controls on Anthropic after they didn't play nicely with the DoD concerning AI use in autonomous warfare. Companies that have hacked other companies and even governments haven't had to suffer any consequences yet.

I suspect this will not change even after 120 days from now.

Vibe Coding Goes Mainstream
For the record, I do not like the term "vibe coding" at all, but it appears we're stuck with it. The term is now mainstream and the concept itself is being used more and more widely. Over the past several days, the internet has lit up over the idea of being able to use apps whose features and design are so closely aligned with Adobe's Creative Cloud that people have wondered if Adobe will sue or at least send a cease and desist order to the makers. And the makers are a lone disgruntled Adobe customer named Brandon Thomas and his trusty AI agent coding assistant, Claude by Anthropic. In what has been reported as being just a few weeks, this duo has churned out most of the Adobe apps and has now moved on to the Microsoft Office Suite and other apps used by creatives such as Avid's Pro Tools, a professional digital audio workstation used by music professionals worldwide. Most of the apps are available to download and install on MacOS, Windows, Linux, and the web for free. Now, to be fair, expert users have installed some of the apps such as the Photoshop and Premiere clones and have declared them to be inferior copies whose functions do not perform perfectly and in some cases, some features are even missing. However, the developer has noted this himself and indicates that he is working on bringing more features to the apps with time. And based on his progress so far, it would seem that not much time would be needed as there are currently 13 apps available on ArtCraft's website with more supposedly on the way. And do not forget that these apps are being offered for free.

The swift creation and deployment of these apps has stirred many conversations across the internet, with some declaring the death of software and others decrying the legality and morality of recreating apps that normally cost money and required many years of research and development to create. Perhaps to his credit, Thomas says that every app was made from scratch using the programming language Rust for broad compatibility across platforms and he did not reverse engineer, decompile, or otherwise copy any of Adobe's code. Instead, he prompted Claude to design and create the apps based the features and structure of the Adobe apps.

Most of the apps are in an alpha state and the code has been released as open-source so that other developers can contribute to the effort. Thomas says his inspiration was his experience with the Adobe suite and his attempts to cancel his subscription, which resulted in cancellation fees he did not appreciate (so say we all). He wanted to prove that with the help of AI, developers can now create apps with the features and polish of the apps from the big software companies that provide alternatives that avoid the expense of recurring fees.

In truth, this approach of creating software with AI assistance is not new, but this is arguably the biggest and most newsworthy example we have seen to date. Most vibe-coded software has been one-off apps for use by individuals to meet personalized needs that commercially-available software was not meeting. Everyone has different ideas of what the ideal to-do, email, project management, note-taking, and personal productivity app should accomplish to be the perfect fit. Being able to create apps with the assistance of AI tools means people can now have apps tailored specifically for them. Some of these apps end up becoming commercially viable and some have ended up in the major app stores for others to benefit. But for the most part, no one had gone after big software houses, especially those with the hefty subscription fees like those from Adobe and Microsoft.

Because Adobe and Microsoft are such big targets, ArtCraft has been a big story hitting all the tech news sites and major YouTube tech channels. But the larger story here is about the future of software development. Once one guy makes some software that competes with professional packages, then it stands to reason that we will likely see more software made with AI assistance that competes with other big name software. Is this a good thing or will it hasten the demise of major software manufacturers? Not long ago, I had a conversation with someone about the likelihood of companies creating their own software and discontinuing their contracts with the major software publishers. Although I acknowledge that some smaller companies or very tech-forward companies may indeed try going down this road, I don't think that corporate software is at major risk right now. One of the advantages of having a relationship with a software manufacturer is that you're assured support and you can assume some level of security. When you are solely responsible for your own software, you take on that support and have to deal with any security issues, not to mention general usability issues. Some companies may be able to handle this extra responsibility, but most will probably continue to pay a fee for the convenience of being able to raise a ticket for someone else to address. Nevertheless, I do think that there will be pressure on software companies and we will see prices come down over time. This will undoubtedly be an issue for the software companies and we will probably see more layoffs or slower hiring as a result.

But how does this translate into software for the everyday consumer? I think the tradeoffs are actually very similar. If you get an AI tool to create a task management app for you, the biggest issue you might have is losing your task data. While that might actually be a major issue for some people, it is generally not a catastrophic issue. But if you lost all your email, that might indeed be catastrophic for many. I would not want to create my own email app and then need to be responsible for maintaining it and troubleshooting it when issues arise. I just want to read and reply to my email. And since many email options out there are free, I don't even necessarily have to pay someone for email. I could very well just choose to have an AI agent help me triage my email or categorize it for me. And when it comes to financial matters, I may find it feels safer to pay for financial management software than try to create my own and risk creating problems that cost me real money or land me in legal trouble. People will continue to use and even pay for professionally-created software on some level. The open questions will be where these lines are drawn at the individual level based on risk tolerance. In aggregate, there will be an impact to the software industry, especially as AI tools become increasingly adept at software development. But most likely we are not all going to create our own suite of apps any time soon.