Technically Working

Michael has been experimenting with Meta Muse, and what caught his attention is not just how it answers questions, but how it can actually go do things. We talk about using AI for reminders, calendar management, travel information, research, and other tasks that usually require jumping between several apps. Michael shares how Muse kept reminding him about a form until he finally finished it, added flights from an itinerary to his calendar, and even noticed emergency contact information that might be useful while traveling. Of course, giving an AI the ability to take action also raises questions about permissions, privacy, security, and how much access we are comfortable giving these tools. Damashe talks about why he prefers limited or read-only access where possible, and we compare the approach with ChatGPT, Claude, Siri, Apple Intelligence, Gemini, and Apple’s Private Cloud Compute. We also dig into Meta’s growing collection of AI hardware, including the Ray-Ban Meta glasses and the Muse Charm, a handheld device designed to provide another way to interact with Muse without wearing AI on your face. Later, Michael explains how he plans to put Muse to another test while attending guide dog school. He wants to use it to help journal the experience each day and see whether a conversational AI assistant can make documenting an entire process easier to stick with. Along the way, we wander through smart glasses, proactive assistants, Tamagotchis, podcasting, and the kind of conversation that started Technically Working in the first place: sitting down, talking about technology, and seeing where the conversation goes. Support Technically Working by contributing to their tip jar: https://tips.pinecast.com/jar/technically-working Find out more at https://technically-working.pinecast.co Send us your feedback online: https://pinecast.com/feedback/technically-working/5bb1b9da-543d-4f62-9f08-235b6ca56e28 Check out our podcast host, Pinecast. Start your own podcast for free with no credit card required. If you decide to upgrade, use coupon code r-431b7d for 40% off for 4 months, and support Technically Working.

Episode Description Michael is getting hands-on with the RØDECaster Pro II, and thanks to Sean’s AccessCaster software, the mixer is finally usable without relying on its touchscreen. Michael and Damashe walk through the hardware, routing, multiple-computer setup, smart pads, monitoring, and what accessible third-party software changes about the value of otherwise inaccessible hardware. The conversation also moves into Michael’s upcoming guide dog training, Apple hardware and the folding iPhone discussion, Siri and podcast transcripts in Earshot, the Technically Working website rebuild, remote mixer control, and Damashe’s experiments using the Drafts MCP connector with Claude. Feedback: [email protected] Website: https://technicallyworking.show Chapters 00:00:00 The RØDECaster Pro II Sounds Different 00:02:23 Getting Started With AccessCaster 00:16:28 Breaking News and Guide Dog Training 00:20:38 Apple Hardware, Macs, and the Folding iPhone 00:23:39 Siri in macOS and Context Menu Changes 00:29:42 Earshot, Siri, and Podcast Transcripts 00:34:03 Going Deeper Into RØDECaster Routing 00:44:32 Accessibility Beyond a Touchscreen 00:49:30 Earshot Codes, Feedback, and Mastodon 00:51:52 Thoughts on Apple’s Latest Event 00:58:34 Rebuilding the Technically Working Website 01:04:51 RØDECaster Wi-Fi and Remote Control 01:11:51 Drafts, MCP, and Letting AI Clean Things Up 01:15:43 Wrapping Up TW180 Support Technically Working by contributing to their tip jar: https://tips.pinecast.com/jar/technically-working Find out more at https://technically-working.pinecast.co Send us your feedback online: https://pinecast.com/feedback/technically-working/d6675ed6-d482-4f7b-b19a-c6736090513d Check out our podcast host, Pinecast. Start your own podcast for free with no credit card required. If you decide to upgrade, use coupon code r-431b7d for 40% off for 4 months, and support Technically Working.

Damashe is joined by Shaun Preece of Double Tap to talk about what happens when AI coding turns “I wish this worked differently” into “maybe I can fix it myself.” Shaun shares how he went from programming years ago to experimenting with Claude Code, while he and Damashe compare how AI has changed the way they research, plan projects, verify information, and learn while they build. They talk about keeping AI from wandering off, asking questions one at a time, using one AI to check another, and knowing when you still need to verify the answer yourself. The conversation gets especially practical when Shaun explains how he used AI-assisted coding to make much of the RODECaster Pro II accessible. That meant figuring out how the mixer communicates with its software, reverse-engineering commands, testing across different setups, and building an interface that can announce changes as controls are adjusted. They also get into computer-use agents, whether AI-generated software has real value, the difference between a quick experiment and a project you actually maintain, and Shaun’s work making Keychron keyboard configuration more accessible. It all comes back to one idea: AI does not have to help you build the next big app. Sometimes it can simply help you fix the things that get in your way and make your computer yours again. Support Technically Working by contributing to their tip jar: https://tips.pinecast.com/jar/technically-working Find out more at https://technically-working.pinecast.co Send us your feedback online: https://pinecast.com/feedback/technically-working/fee388ee-c70a-4fb0-a6e0-cc3de959c4f3 Check out our podcast host, Pinecast. Start your own podcast for free with no credit card required. If you decide to upgrade, use coupon code r-431b7d for 40% off for 4 months, and support Technically Working.

Technically Working 178: I Just Don’t Want to Do It Show Notes Michael and Damashe start out talking about GPT 6 and the ways AI has moved beyond being something interesting to experiment with and into something they actually use to get work done. Damashe shares how ChatGPT helped him troubleshoot a machine by finding and working through the documentation, while Michael talks about using computer use to handle WordPress changes and other tasks he already knows how to do but, as he puts it, sometimes just doesn't want to do himself. That leads into a conversation about when handing work to AI makes sense, when you still need to check its work, and how Michael is limiting access when letting AI work on websites. They also talk about volunteer work, last-minute requests, and Michael's goal of being more selective about the unpaid work he takes on. Then Michael asks for "no judgment" before revealing his latest purchase: a RØDECaster Pro II. His current mixer is starting to show its age, and the RØDECaster offers dual USB connections, mix-minus, multiple outputs, and the flexibility he wants for live and hybrid audio work. The problem, of course, is accessibility. Michael and Damashe talk through whether the RØDECaster can be made usable through existing workarounds, custom software, computer use, or direct device control. From there, the AI conversation returns because apparently every website needs an AI now. They consider whether smaller local models could actually be useful inside apps for things like better podcast search without turning every product into another AI chatbot. With an Apple event approaching, Michael and Damashe make some predictions. A folding iPhone leads to theories about Touch ID, including Damashe's hope that Apple puts it in the Apple logo on the back. They also talk Apple Watch, HomePod, Apple TV, Siri, what they might buy, and how Apple could handle a major event and a future leadership transition. Michael has also been relying on Siri more than he expected. Some recent successes with messages, calendar events, and reminders have him wondering how useful Siri could become once third-party apps can expose more of their features. That naturally turns into what Siri support should look like in a podcast app. Damashe wants simple commands like playing the latest episode of a show, continuing the queue, or moving to the next item. Michael shares some of the Siri work planned for Earshot along with recent player improvements. They wrap up with more Earshot changes, including easier episode and chapter navigation, folders that survive OPML exports, episode filtering, and finally getting auto-expiring podcasts working. In This Episode Using GPT 6 for research, troubleshooting, planning, and everyday work A free month of Earshot Plus for Technically Working listeners Letting computer use handle the tasks you know how to do but don't want to do Reviewing AI work instead of blindly trusting it Limiting AI access to WordPress sites Cutting back on last-minute and unpaid work Michael buys a RØDECaster Pro II The accessibility problem with modern audio hardware Could custom software make the RØDECaster accessible? AI showing up everywhere Where smaller local AI models might actually be useful Folding iPhone and Touch ID predictions Apple Watch, HomePod, Apple TV, and Siri expectations Michael starts trusting Siri a little more What Siri support should look like in a podcast app New Earshot playback and navigation controls Preserving folders in OPML exports Episode filtering and auto-expiring podcasts Chapters 00:00 - GPT 6, Earshot Plus, and putting AI to work 05:00 - ChatGPT versus the technician 09:50 - Computer use and "I just don't want to do it" 12:51 - Doing less free work 13:54 - No judgment: Michael bought a RØDECaster Pro II 20:00 - Can inaccessible audio gear be made accessible? 26:31 - Everybody has an AI 28:36 - Apple event predictions 34:20 - What Apple hardware are we actually buying? 40:00 - Siri, Apple events, and what comes next 50:00 - Michael is starting to rely on Siri 53:30 - Calendars, reminders, and task apps 56:45 - What should Siri do in a podcast app? 60:00 - New Earshot playback controls 63:20 - OPML folders and different ways of listening to podcasts 65:18 - Filters, auto-expiring podcasts, and finally solving the original problem Links Mentioned Technically Working Feedback: [email protected] Support Technically Working Earshot Podcasts on the App Store Support Technically Working by contributing to their tip jar: https://tips.pinecast.com/jar/technically-working Find out more at https://technically-working.pinecast.co Send us your feedback online: https://pinecast.com/feedback/technically-working/5b9d0997-8857-41a0-a909-e2dd90002cde Check out our podcast host, Pinecast. Start your own podcast for free with no credit card required. If you decide to upgrade, use coupon code r-431b7d for 40% off for 4 months, and support Technically Working.

Technically Working 177: The Bot Couldn’t Get Out Show Notes Michael and Damashe begin with a look at how AI can turn convention documents into WordPress pages, Gravity Forms, confirmation messages, and other ready-to-use materials. They discuss why a smoother payment process matters and how showing people only the instructions they need can prevent abandoned registrations and unpaid entries. Then Damashe reacts to rising Mac prices, explains why selling his MacBook Pro may have been a poorly timed decision, and considers whether things have become bad enough to buy another Windows computer. Michael shares useful shortcuts and recent experiences with Codex and ChatGPT, while both hosts stress that AI-generated work still needs to be reviewed. The conversation also covers Postmark, Cloudflare, transactional email, and the decision to rebuild the Technically Working website with familiar WordPress tools. Damashe explains why he does not want his business depending on a software stack he barely understands. Along the way, the hosts offer a warning about installing several WordPress caching plugins and explain why more caching does not make a website faster. Michael shares the latest Earshot improvements, including faster feed refreshes, haptic feedback, adjustable descriptions, new accessibility settings, and category browsing. The two consider how server-side podcast feed checking might work and what it could cost. Finally, Damashe explains why the Technically Working Mastodon bot never appeared during the ACB convention. He secured the server so thoroughly that the bot could not get out. The planned solution involves a Raspberry Pi, Tailscale, and one more project competing for his attention. In This Episode Using AI to build WordPress pages and Gravity Forms from existing documents Creating clearer registration and payment flows Why AI-generated forms and websites still need human review Rising Mac prices and Damashe’s poorly timed MacBook Pro sale ChatGPT and Codex shortcuts and computer control Postmark, AWS SES, and Cloudflare for transactional email Rebuilding the Technically Working website with GeneratePress Choosing a software stack you understand well enough to maintain Why several WordPress caching plugins will not make a site faster Earshot feed refresh, haptics, descriptions, and category browsing The possibilities and costs of server-side podcast feed checking Panic returning money to Playdate customers The mystery of subscriber number one How Damashe accidentally trapped the Technically Working bot Tailscale, Raspberry Pi projects, cameras, and internet failover Support Technically Working by contributing to their tip jar: https://tips.pinecast.com/jar/technically-working Find out more at https://technically-working.pinecast.co Send us your feedback online: https://pinecast.com/feedback/technically-working/ba76eb96-515b-4c08-9587-0684eae26143 Check out our podcast host, Pinecast. Start your own podcast for free with no credit card required. If you decide to upgrade, use coupon code r-431b7d for 40% off for 4 months, and support Technically Working.