We also spent time talking through the practical realities and failures of current AI tools, including Mike's experience using Microsoft Copilot for employee performance reviews and how it struggles to trace back to its original email sources. From there, our conversation turned to the evolving mobile hardware landscape. We compared our current daily devices—including Mike's experience with the Samsung Galaxy S24 Plus and Galaxy Tab S11 Ultra—and weighed in on the upcoming hardware trends, from memory constraints forcing developers back toward code efficiency to what we hope to see from Apple's eventual entrance into the foldable market.

It was a great, wide-ranging discussion covering everything from 6502 assembly language programming on the Apple II to modern Gemini Spark agents automating my daily tech news briefs. 

We wrapped up with a promise not to wait another decade before the next chat, with plans to reconnect around Canadian Thanksgiving in October. Check out the full episode video and transcript to hear all the details!

I noted my initial experience using Google Spark, an autonomous personal AI agent that recently rolled out to Google AI Pro subscribers. This capability was once restricted to the much more expensive Google AI Ultra tier. In my test, I instructed Spark to review the extensive archive of past MobileViews Podcast show notes alongside current technology news platforms. Utilizing the recurring themes from our previous episodes, the agent generates a compiled tech news digest daily at 6:00 PM US Eastern Time, aligning with the conclusion of the standard workday.

A major topic was Jason’s journey into "vibe coding". While initially hesitant, he used AI to build a custom Tampermonkey script that saved his team over 150,000 manual clicks by streamlining a web form. He later used AI to develop a Mac OS tool called "File Time Travel" to fix frustrating video timestamp metadata issues in Google Photos—a highly specific problem that software developers have not addressed. We discussed how AI flips the traditional learning curve on its head, empowering non-developers to create deeply personal, single-purpose software solutions.

We wrapped up by exploring the power of voice-as-input for AI. Jason has been using Whisper Flow, an LLM-powered speech-to-text tool that finally fulfills the broken promises of old software like Dragon Dictate by actually understanding conversational context and complex names. This allows for "ramble coding"—speaking massive, context-rich prompts directly into the AI to generate much better code and outputs. Of course, the AI revolution isn't without its hilariously frustrating bumps; we swapped stories about AI hallucinations, from Microsoft Copilot completely fabricating Yellowstone episode descriptions.

The core of Jon’s summer project was migrating away from the Day One journaling app to avoid its $25 yearly fee and proprietary cloud storage. Using ChatGPT and Codex, he generated scripts to convert his Day One JSON export into future-proof Markdown files managed within an Obsidian vault. He then successfully replicated Day One's best features, using Apple Shortcuts and Python to ingest text snippets, process daily photos, perform offline audio transcriptions, and even selectively transcode video files larger than 25MB down to mobile-friendly sizes.

Expanding his DIY software suite, Jon also automated his personal relationship management and location tracking. He built a script that "interviews" him weekly to automatically update his self-hosted Monica CRM and Obsidian vault with details about his interactions with family and friends. Furthermore, to reclaim his travel history after Google restricted the web version of Google Maps Timeline, Jon coded a tool to parse his device's local JSON location data into detailed, daily Markdown travel logs. With his self-hosted documentation ecosystem fully functional, Jon is taking it on the road for late-summer travel and will return to the podcast in mid-August.

The core of our conversation focused on the profound impact AI is having on higher education and course design. Jon noted that as AI companies merge their tools—such as integrating Codex directly into the ChatGPT Mac app—automation is becoming much more holistic. He detailed his new workflow for building his clinical psychology and statistics courses, using AI to spot curriculum blind spots, draft syllabi, and suggest relevant journal articles. Most impressively, he now uses Codex to automate the tedious administrative "grunt work" of building modules and formatting links directly inside the Canvas learning management system, freeing him up to focus on transitioning his classes toward project-based learning rather than traditional exams.

We wrapped up the episode with a discussion on the psychological trap of having a dedicated "tech budget". We agreed that it is too easy to buy devices just because the funds are available, leading to cognitive dissonance when you feel forced to use hardware you don't actually need. Instead of buying minor upgrades, Jon is actively saving his budget for the rumored foldable iPhone that might be announced later this year.

I shared my frustrations with Microsoft Copilot, which is rapidly burning through my AI credits just for using basic OneNote features like syncing handwritten notes. This has pushed me to experiment with Obsidian and the Excalidraw plugin as a potential offline alternative for my e-ink tablet. We also touched on Windows, discussing the new Cloud Rebuild feature in the Windows 11 preview. Don shared his success in reviving an older 2019 Samsung Galaxy Book with a simple Windows 11 reset, a timely win given that skyrocketing RAM and SSD prices continue to make budget-friendly options like the MacBook Neo incredibly appealing.

Finally, Don provided a deep dive into the complex world of Microsoft Enterprise Licensing. He explained how Microsoft is rolling features from its struggling Intune suite into its standard E3 and E5 licenses while simultaneously pushing price increases. Most notably, they have introduced a new E7 license—which Don jokingly called the "mother of all licenses"—that includes high-end Copilot agent features for a staggering $99 per user per month. We closed out with a look at how these price hikes, including those for non-profit organizations, are forcing IT departments to carefully rethink how they assign user licenses across their organizations.

Jon, meanwhile, has been putting two impressive new hardware gadgets through their paces. He’s been testing the Insta360 Luna Ultra, a gimbal camera featuring a clever detachable screen that doubles as a remote control, allowing him to film himself while managing tracking and settings from a distance. He also showcased the Kodak Zink Slim Printer, a pocket-sized device that creates color adhesive stickers using zero-ink, heat-sensitive paper—perfect for adding instant color photos to a physical journal. While these gadgets provide plenty of utility, the broader tech landscape is facing a sobering shift: AI-driven demand for components is causing shortages in RAM and SSDs, leading to significant price hikes for devices like the MacBook Neo, which is expected to jump in price from $599 to $699.

Jon shared a major shift in his productivity workflow by moving to Obsidian, a "Swiss Army knife" of note-taking. By using Codex to convert 20 years of WordPress entries and Day One journals into Markdown files, he has created a future-proof, portable "vault" that avoids proprietary databases. We also discussed the release of Android 17, which introduced an interesting "Screen Reactions" overlay feature but also caused frustration by resetting permissions for tablet casting and photo galleries.

To wrap up, Jon provided a field report on his DJI Neo 2 drone, which successfully tracked him during a 20mph e-bike ride. Despite suffering its first high-speed crash into a tree, the lightweight drone proved remarkably durable, surviving the impact with no visible damage. We also touched on a few tech trends, including Gen Z's growing rejection of Silicon Valley's vision in favor of "dumb" tech like flip phones and repaired iPods

A major portion of the episode was dedicated to my "credit crunch" rant regarding Microsoft Copilot. I discovered that Microsoft’s 365 family plan only provides 60 AI credits per month, and the "intentional use" policy is aggressive. According to Copilot itself, credits can be consumed simply by opening the app, syncing handwriting from an e-ink tablet to OneNote, or even having the AI suggest a grammar fix you don't actually use. This led me to explore Obsidian as a OneNote alternative, as it offers free handwriting plugins without the credit overhead. Jon suggested a sustainable path forward: using AI to build offline scripts or tools that perform data manipulation locally to avoid recurring token costs.We also looked at the hardware horizon, specifically Microsoft’s announcement of Project Solera—AI-powered badges and desktop displays—and the new Nvidia RTX Spark PCs,. These machines are purpose-built for local AI, boasting a petaflop of performance to run personal agents offline. Finally, with Apple WWDC just around the corner, we shared our hopes for the long-promised "personal context" updates to Siri. Jon is also eagerly awaiting his pre-ordered Clicks communicator and keyboard, while I continue to hold out hope for a MacBook Neo with a backlit keyboard and a desktop Mac Neo. Whether it’s navigating "vibe coding" loops or managing AI budgets, it’s clear that the "magic math" of the AI industry is starting to meet the reality of the bean counters.

A major theme of this episode was the shifting economic reality of AI. Don, ever the healthy skeptic, compared the current AI hype to the dot-com bust and the "magic math" of non-GAAP reporting. We discussed the "tokens vs. humans" trade-off, noting that as companies like Google and OpenAI move toward token-based pricing, the cost of farming out thinking to AI may soon exceed the cost of hiring a human. I’ve been finding ways to outsmart these limits by using standard LLMs to "interview" me and generate highly efficient Codex prompts, which usually get the job done on the first try.

We also tackled some serious infrastructure and software headaches. I’m currently dealing with expiring Secure Boot certificates on some of my no-name PC boxes—a situation Don rightly called a "mismanaged" disaster on Microsoft's part, given the lack of clear documentation and the bugs causing BitLocker prompts. On a more positive note, my 8GB MacBook Neo continues to impress me with its efficiency, proving that tight optimization can often trump raw specs. We wrapped up with a look at the future of Nvidia ARM-based processors for Windows and a strange sighting during my daily walk: a fiber optic cable hanging at neck height between utility poles, a reminder that even high-tech infrastructure can have very physical (and dangerous) failures

On the software and AI front, we looked ahead to Android 17 and its new "Rambler" feature for Gboard, which uses AI to filter out "ums" and "ahs" from voice transcriptions. I shared my experience with Gemini Pro’s voice cloning, which was "scary good" at mimicking my voice with minimal training, while Frank voiced skepticism about Wear OS 7 replacing tiles with widgets, fearing it's a step backward for round-screen usability. The episode also served as a warning about the dangers of auto-updates; Jon shared how a Ubiquiti router update broke his HomeKit setup—requiring an SSH command to fix—and I recounted a corrupt OneDrive for Mac update that forced me to roll back to its previous version using Time Machine.

We wrapped up with a demonstration from Sven, who demonstrated his new Pixel Fold 10's unique feature in Google Meet that allows for a split-view using the front and back cameras simultaneously. This "double-vision" mode allows a caller to show their face while also providing a high-resolution view of their surroundings, which we agreed would be a game-changer for remote tech support or traveling. d there is still plenty of innovation happening in the Android ecosystem.

The second half of our conversation turned toward the cognitive impact of AI. We debated a recent study claiming that using AI for even ten minutes can impair brain performance, particularly in solving math problems. Jon Westfall, drawing on his background as a cognitive scientist, was skeptical of the study’s protocol, while I noted the striking similarity to the "brain mush" warnings I heard about calculators fifty years ago. We wrapped up by discussing how to "smooth over" the friction in our personal hobbies. For Jon, this means a "buy once, cry once" philosophy with high-quality tools like professional knitting needles. For me, retirement has brought a surprising fascination with the mechanics of pencil sharpeners, where I’ve discovered that investing in a high-quality long-point sharpener is often the key to enjoying the simple act of doodling on paper. Whether it's using AI to remove "ums" and "ahs" from speech via Gboard’s new Rambler feature or whittling a pencil to perfection, we agreed that the right tool fundamentally changes the experience

Jon and I also explore the practical and social implications of AI in the workplace, particularly focusing on an article I found about the trend of "whispering" to AI instead of typing happening in offices. Jon shares how rambling to ChatGPT's voice mode for five minutes helps him rapidly prototype structured op-ed pieces, essentially using it as a highly effective sounding board and editor. However, we both point out that this shift toward voice interactions makes open-plan offices—an environment I already strongly dislike—even more distracting, making modern workspaces resemble call centers. To round out the episode, we tackle a few persistent societal myths. Jon draws on his background teaching educational psychology to debunk the concept of preferred learning styles, and I acknowledge that the popular 10,000 steps a day health goal is largely an outdated pedometer marketing gimmick, even though I still try to hit that target to stay active.

Jon shared a surprising shift in his workflow: returning to paper notebooks during meetings to improve interpersonal engagement, while using a Plaud Note device for the actual transcription. He also showcased his latest "vibe coding" success, building a card game app called "Pierogi Palace" in just 48 hours using ChatGPT and Codex to handle everything from the rule set to App Store screenshots. Meanwhile, I offered a mini-rant about Microsoft 365 Copilot, which has turned the once-simple task of scanning my daily doodles into a frustrating five-click process compared to the old, streamlined Microsoft Lens app.

Looking ahead to Google I/O, we discussed the rumored "Aluminium OS," an AI-first operating system that aims to merge Chrome OS and Android. We wrapped up with some tech nostalgia, reflecting on the days when the original iPhone and Windows Phone 7 lacked built-in screenshot features—a sharp contrast to today's AI tools that can generate entire apps in a weekend. I also gave a shout-out to my 2019 Anker Bluetooth keyboard, which still works perfectly for my iPad Mini typing needs, proving that some hardware really is built to last.

I spent some time on a "mini rant" about the state of budget Android tablets; while you can find great hardware for under $150, these devices almost never receive security or OS updates, effectively making them disposable e-waste. On the infrastructure side, the "router apocalypse" continues as the FCC's ban on foreign-made hardware looms, though Amazon Eero recently secured an exemption through October 2027. To prepare for a post-exemption world, I’m preparing to experiment with OpenWrt on an old travel router.

We wrapped up with a look at the startling 2026 price hikes for storage—where a SSD drive that cost $350 last year is now nearly $1,000—and a discussion of encouraging  maker culture for young people. We debated whether encouraging preteens to use 3D printers and AI coding tools like Codex is better than traditional hobbies, reflecting on the Heathkits and "dangerous" chemistry sets of my own youth.

We also tackled the 8GB RAM debate, finding that the MacBook Neo performs remarkably well for most users because Apple optimizes its OS for its hardware so efficiently. We compared this to the increasingly bloated experience of Windows 11, which forces users to have 16GB or more of RAM. Jon shared his experience upgrading to Google One AI Pro to leverage NotebookLM for his university courses, transforming long lectures into concise, AI-generated study materials. He even demonstrated the value of paid AI tiers by building a custom Strength Tempo Trainer web app in just 20 minutes using Codex. We discussed the "AI divide" comparing free AI tiers to paid ones. Finally, I’ve found Alexa Plus surprisingly useful for troubleshooting Amazon specific and Echo device glitches through its browser-based interface.

I have become a massive fan of NotebookLM and the new notebooks feature within Google Gemini,. I recently fed hundreds of our podcast show notes and blog posts into a Gemini notebook to see how it handled the data. We discussed how this technology is becoming popular in academia as a tool for students to engage with material, though it poses a significant threat to the textbook industry by easily creating the same supplemental materials publishers charge for,. I’ve even been using it to build animated video presentations and conduct additional research, which are features included in the $20-a-month AI Pro subscription,.

Our conversation turned toward the concept of "digital sovereignty," a movement in the European Union to reduce dependency on U.S. tech giants. We looked into the history of open-source office suites, from StarOffice to the current fragmentation of LibreOffice,. Interestingly, the EU is now looking toward "EuroOffice," a fork of OnlyOffice (which itself has roots in Latvian and Russian development) to replace Microsoft Office. Steve, Jon, and I debated whether this would be a cleaner transition than past attempts, especially as corporate customers grow increasingly annoyed with the "nickel and diming" of subscription services,.

Jon and I revisited the "technology gap" we see in students who have grown up exclusively on Chromebooks. Many struggle with the basic concept of a file structure or how to actually download a file to a specific directory. My own daughter recently joked that Windows is for "old people," which signals a shift Microsoft should be wary of. However, Jon shared a positive note on the democratization of app development. He built a purpose-built fitness timer for his strength training in just 20 minutes using AI,. While this "one-two punch" of AI interviewing and coding is powerful, I cautioned against the "Microsoft Access problem"—the risk of non-programmers building mission-critical tools that lack documentation or error-checking,.

On the hardware front, Jon shared his first impressions of the MCON controller, a slider-style mobile gaming device that features a built-in MagSafe stand. While the buttons are a bit small for some, its "pocketability" makes it a strong contender for travel,. Steve updated us on the Sea Otter Classic, essentially the "CES for bikes," highlighting new e-bike motors that are rejuvenating the industry,. We also touched on the EV market, specifically the Scout SUV and its move toward a range-extending motor. Steve even noted a great practical tip: using an EV as a reverse-load power source to keep a refrigerator running for up to two weeks during a blackout.

Listen to the full podcast to hear Steve's story about meeting will.i.am at CES.

I’ve been getting a lot of mileage out of Google AI Pro, and they recently gave their subscription service more value by bumping the storage from 2TB to 5TB for the same $200-a-year price. One of the most impressive new features is NotebookLM’s cinematic video creation, which I used to analyze and summarize 18 years of our own podcast history. We also discussed Google Flex OS, which now offers a three-dollar bootable USB stick for people who want a low-barrier way to try Chrome OS on their existing hardware.

We shared our "origin stories" with Apple, and it was a fun look back at how the technology evolved:.

I also recounted my favorite customer service story from 2010, where an email to Steve Jobs' office resulted in me getting a launch-day iPad despite a UPS delivery delay—and UPS actually made its first-ever Saturday delivery in my region just to get it to me and other iPad pre-order customers..

We revisited the MacBook Neo. While critics focus on the 8GB RAM limit, the build quality is miles ahead of the "plasticky, bendy" Chromebooks that dominate the budget market. Jon noted that his students are the real target for this device; they want to be in the Apple ecosystem but often can't afford the Pro models, making the $499 education price a potential game-changer.

We closed out the episode with discussions of the WiFiPorter Sven bought to provide guests to his home with a seamless guest WiFi connection and the MCON magnetic transforming game controller that Jon recently purchased.

What I found truly remarkable was that Jon was able to build a working model in only 90 minutes. He used what he calls a "one-two punch" with AI tools:

• The Interview: He first used regular ChatGPT to "interview" him about his specific requirements and ideas.
• The Build: Once the requirements were fleshed out, he had the AI write a high-quality prompt for OpenAI Codex, which then built the actual application using TypeScript and React.

The project is currently hosted on GitHub Pages, which Jon set up so that the site automatically rebuilds and deploys in about a minute every time he pushes a change to his repository. To make the tool more accessible, he included real-world examples, such as independent t-tests for tutoring programs and chi-squared independence tests for marketing surveys