Podcast
James' Audiolog: Indie.am
Hosted by Indie.am · EN
Hey, I'm James, and welcome to my personal Audiolog. I mostly use this to share personal updates, via Indie.am. DM me for an invite!
14episodes
Episodes
Newest firstAll episodes
Waitlist SaaS: Good Idea, or Bad?
3w agoTap to summarizeIs It a Good Idea to Launch a Waitlist SaaS? What is a Waitlist SaaS? A Waitlist SaaS is a web application designed to simplify the process for founders to add a waitlist sign-up to their product launch strategy. It sounds appealing, but is it really a good idea? Why It Might Not Be a Good Idea Background The concept of using a waitlist as a launch strategy was popularized by companies like Robinhood and Clubhouse. These companies created hype around their products by making sign-ups initially impossible, allowing only a waitlist. Users could invite friends or share on social media to move up the list. This strategy became popular for SaaS launches following these examples. The Reality Behind Successful Launches However, the success of Robinhood and Clubhouse wasn't solely due to their waitlists. Their real strategy was leveraging FOMO (Fear of Missing Out) and hype. They over-promised on their products, creating a buzz that compelled users to share the product to move up the waitlist. The waitlist added friction to the sign-up process, which could actually hinder conversion rates. Conversion Rates and Hype Consider this: 100% of visitors willing to sign up for a waitlist would likely have signed up for the product immediately if it were available. The waitlist conversion rates—whether a week, a month, or two months later—cannot exceed 100%. Would removing the waitlist have slowed down the hype and FOMO driving the marketing campaign? Possibly. Conclusion In conclusion, launching a Waitlist SaaS in 2026 may not be a wise idea, especially for solo founders or small teams. While hype and FOMO can be effective launch strategies, they require a level of execution typically found in more experienced teams. A SaaS that facilitates adding a waitlist sign-up process is not likely to be a mass-market strategy that inexperienced or small teams can successfully implement. These elements are probably the least challenging parts of a more complex marketing strategy that a skilled marketing team might employ.
Transcribe →
Did the AI bubble just Burst?
Jan 28, 2025Tap to summarizeDid the AI Bubble Just Pop? Hey, so the question on everybody's mind today is, did the AI bubble just pop? I think no. The stock market thinks yes. Let's explore why I believe that's a bit short-sighted and not really considering the big picture. AI in 2025: A New Frontier Here's how I frame it: the things I use AI for in 2025 that weren't possible 24 months ago are remarkable. I use AI to: Write and edit blog posts Create music Generate fan art Summarize YouTube videos Transcribe audio and video Assist with language learning Replace Google searches in many cases Write database queries, especially when not using SQL Perform back-of-the-envelope calculations Translate manga Code and build apps, just like this one Two years ago, I didn't use AI for any of these tasks because it simply wasn't possible. The models weren't trained yet; they didn't even exist. We're only a couple of years into this massive transformation, and the potential applications of AI are just beginning to be realized. The Untapped Potential of AI There are still areas where AI hasn't made its mark. For instance, I can't use AI to do my taxes, and there's no AI-powered legal advisor available. The training data is all there—case law is just 150 books on someone's bookshelf that someone has spent a decade reading and understanding. I'm not suggesting AI could take over the legal industry, but it seems plausible that knowledge of tax law or other legal fields could be made more accessible through a well-trained AI model. However, that's not available yet. The Demand for AI Remains Strong The demand for AI hasn't changed. The headline claims that a model has been trained for $6 million to compete with OpenAI's cutting-edge model are, in my view, largely propaganda. Significant advancements have been made, such as each token activating fewer parameters and the use of a mixture of experts. I believe that soon, OpenAI will disclose how they made GPT-4 Turbo cheaper and more resource-efficient than previous versions of GPT-4. We've been working on optimizing these models for years, even before some companies started, and we were actually ahead in this area. Infrastructure Challenges Building a data center to run these models still requires hundreds of millions of dollars. It's not like DeepSeek is suddenly going to run on a different chip. They will still operate on H100 in data centers, using eight-user servers that consume vast amounts of electricity and human resources to build, deploy, and operate at hyperscale. We're still not meeting the demand for AI capacity. More people want to use tools like Stable Diffusion or ChatGPT than there is capacity for right now. There are still new applications for the API, regardless of the model you choose. The DeepSeek Story That's another part of the DeepSeek story. How do you make the claim that you trained a model for $6 million on half a billion dollars worth of GPUs that are required? The story continues to unfold, and the potential for AI remains vast and largely untapped.
Transcribe →
New feature: Logo Idents!
Jan 25, 2025Tap to summarizeIndie Maker Update Hey there, welcome back. I'm James, and this is my indie maker update for Saturday. New Audio File Format What you're hearing is a new audio file format. Optionally, you can add a logo identity to the beginning of your personal podcast updates. This addition makes it sound a little bit more professional, a little bit more like a real podcast. However, right now, it's limited to one very generic, very inoffensive logo identity. Logo Identity I handpicked and edited the logo identity to be short and very easy to add to any spoken audio track without requiring much tweaking to the way you record your audio. It works best if you start your audio in a certain way, like "Hey, welcome back" or "Hey, I'm James," instead of the more typical, boring way I usually start my audio recordings. Conclusion Hopefully, if this test goes well, this will be the track that is published and introduces the new feature. Thanks for listening.
Transcribe →
Indie.am Update: Improving Audio Quality
Jan 8, 2025Tap to summarizeIndie Maker Update: Audio Processing Improvements Second Indie Maker Update of the Day Aren't you lucky? Today, I spent about four hours fine-tuning my FFMPEG pipeline. Here's a rundown of what I worked on: Enhancing Audio Quality I experimented with a recurrent neural network to denoise the track. Additionally, I aimed to improve the audio quality by disabling certain audio helpers in the getUserMedia request. These helpers, such as echo suppression, noise suppression, and level normalization, are often applied by iOS or Android browsers. While they help in some cases, they can degrade the audio to something akin to telephone quality. By turning these off, I managed to elevate the quality to something closer to native recorder medium settings. Pipeline Adjustments I made several tweaks to the pipeline that converts audio to MP3: EQ Modifications: I applied a few equalizer adjustments to enhance the sound. Level Normalization and Denoising: I normalized the audio levels and applied denoising techniques. Band-Pass Filter: Implemented a simple band-pass filter to eliminate electrical hum. Click Removal: Used plugins to remove clicks from the audio. Silence Removal: Added a plugin to eliminate dead noise zones during periods of silence. Results and Next Steps Overall, I believe I've significantly improved the audio quality compared to a few hours ago. I'm quite pleased with the results and look forward to testing how it performs on other devices tomorrow.
Transcribe →
How does Google Scrape library data from Itunes?
Jan 6, 2025Tap to summarizeGoogle's Intriguing Access to iTunes Music Library Hello! Today, I want to discuss something intriguing that caught my attention a couple of months ago and has been on my mind ever since. It appears that Google has found a way to scrape data from an iTunes music library on the same device through one of their apps. However, I'm uncertain which app is responsible, and it doesn't seem to involve any information sharing or consensual data sharing with Google. The Mystery of Unrelated Music Recommendations I have a specific set of music, primarily tracks I copied into my iTunes music library years ago, which are now part of my Apple Music library. These tracks have never left the app, and I don't search for these artists or songs on YouTube. Interestingly, this niche music from my library began appearing in YouTube notifications and recommendations around November or December, seemingly out of nowhere. Unexpected YouTube Music Suggestions YouTube started suggesting, "Hey, you don't listen to music on YouTube. Why don't you try YouTube Music? We already have all of your music from your library. You could switch right now and start listening." This was surprising because my music was siloed in iTunes and Apple Music. I'm unaware of how Google could access it without some OAuth mechanism. Potential Data Access Methods In iOS app development, there's a mechanism to add a "Sign in with Apple Music" button, allowing third-party apps to access data. However, it seems Google has found a way to scrape song names and artist names from the entire iTunes music library, possibly through one of their installed apps on my device. This data appears to be used to promote YouTube Music, a competitor app. Apple's Potential Concerns It's puzzling because one would assume Apple would go to great lengths to prevent a competitor app from scraping this data. Presumably, the only use for this data is to advertise within Google apps, encouraging users to migrate from Apple Music to YouTube Music. It's surprising that Google has managed to achieve this, despite Apple's likely efforts to prevent it. Distinguishing from Common Misconceptions Before jumping to conclusions, I've thought about this extensively. This isn't like those instances where people Google something and then see related ads elsewhere, not understanding how tracking works. I'm aware that phones aren't spying on us in that manner. While this situation might sound similar, I've considered it for months and can't think of any way Google could access this information except through an app installed on the same device. Possible Methods and Speculations Perhaps it's easier on macOS, and maybe Google used the Chrome binary to read my iTunes music library, which would be even more surprising. Whatever method they employed to obtain this data is likely interesting, and I imagine there are some Apple developers sending angry emails to Google about not adhering to developer terms. In conclusion, I find this situation fascinating and worth pondering. How Google managed to access this data remains a mystery, but it certainly raises questions about data privacy and app permissions.
Transcribe →
Indie.am Updates: RSS fixes, social sharing options
Jan 6, 2025Tap to summarizeIndie.am Update for Monday Recent Fixes and Improvements RSS Feed Fixes I've just published some fixes to the RSS feed. Years ago, I hardcoded 2021 as the year, which caused issues with all subsequent feeds. I've now corrected this mistake. Additionally, I'm using ChatGPT to format transcripts into blog posts, converting Markdown to HTML. This ensures that the content is nicely formatted when opened in podcast apps. Custom Component for Podcast Apps I've also worked on a custom component to integrate the old Podlog. This feature allows users to click an RSS icon in the app, redirecting them to their chosen podcast app based on their platform. It's a small but significant fix. Background Work Transcriptions with Whisper Transcriptions are now handled by Whisper, which provides high-quality results. However, it's resource-intensive on the server. I've implemented a system that queues jobs as users browse older Indie.am podcasts, re-transcribing them with Whisper for more accurate timestamps, especially for subtitles. Video Generation I've reworked video generation but haven't integrated it into the apps yet. This feature will be complex, involving 15 video templates. Users will be able to: Add a share button Create a video Choose a template from the gallery Customize options The video generation process will take about 10 to 20 seconds. If you exit, it will save back to the item and notify you once complete. This complex UI feature is still in development. Social Media Sharing Enhancements I've made several improvements around social media sharing, including meta tags and adding a share button. This is essential for sharing content to other social media feeds. You can embed it in Twitter (formerly known as X), but Blue Sky doesn't yet support audio posts. You can link to them, but playback issues occur due to their in-app browser. Blue Sky Playback Issue There's a peculiar issue with Blue Sky's in-app browser muting audio by default. If I share an audio post to Blue Sky and launch the podcast app via their browser, I can then return and play the audio file. It's a strange behavior, and I'm investigating why this happens. Conclusion I've accomplished a lot and thought it would be fun to share these updates. I'm also testing out the long recording feature. I'll stop here and see how this turns out.
Transcribe →
Indie Maker Update
Jan 6, 2025Tap to summarizeIndie Maker Update: Final Thoughts on Indie.am All right, this is probably my last Indie Maker update specifically for Indie.am. This will be my Indie Maker update in general then. Completion of Indie.am Features I've basically completed all the features I wanted to complete for Indie.am. Most of this is around GPT and video generation, and those things are working fairly well. I'll continue to use the app and make little tweaks, but I'm pretty happy with where it is. This is more of one of those baseline things that I'm really happy that it exists again. I don't have to go through TestFlight, and I want it just so I can publish my own audio blog and see how it turns out. New Creative Pursuits Lately, I've been focusing on a different app to stream music and working on some different creative pursuits. I've been writing music, recording, and streaming a couple of games on YouTube, which has been a lot of fun. I'm trying to get in the mindset of creating content in different mediums, playing around with it, and seeing what I can learn from that. I'm having a lot of fun. Development Insights I've got like four or five of these Capacitor Framework 7 Vue apps, and each one I build feels a little bit more native and polished. I'm going back and forth between them, trying to re-implement the things I've learned from one app to the other and solving some issues. For the PWA, I basically have to swap out Framework 7's Vue router for Vue Router to use Vue with Framework 7 in a way that works and feels good. I had to figure out how to get all the animations and swipe back functionality to work. I also usually swap out Framework 7 Store for Pinia and have moved away from any Capacitor APIs unless absolutely necessary. This allows me to build the PWA straight from Vue and only use Capacitor where I need the extra functionality. Building a PWA in Vue with Framework 7 versus a Capacitor build of that PWA results in a completely different set of little works to the app. Future Plans I need to go back and update the Mastodon client that I created. I want to get short video working for when TikTok shuts down in a couple of days. I have a TikTok-style video feed inside of that app, but I need to swap out Vue Router for a Framework 7 Router to fix a few bugs and glitches. Once that's done, I can put it back up. Other than that, I've mostly been focusing on music right now. Maybe I'll share more of that in the next few days, but as far as my Indie Maker update goes, that's pretty much it. Thanks for listening.
Transcribe →
US Steel and Aluminum Tariffs
Jan 6, 2025Tap to summarizeUS Steel and Aluminum Tariffs: A Rookie Mistake Hello, everyone. Today's discussion centers on the US steel and aluminum tariffs. I must say, this seems like a real rookie mistake. Imposing tariffs on inputs rather than outputs is a flawed strategy. US consumers don't directly purchase steel or aluminum, nor the byproducts of imported steel and aluminum. Instead, they buy goods whose value has been increased multiple times by various companies after the initial input was taxed. Learning from the Past We've been down this road before. In 2018, the US imposed similar tariffs on the same countries. We have case studies available that detail the effects of those tariffs. What happened was that US producers overbuilt capacity, leading to a significant bubble. Demand only rose modestly, covering just a fraction of the additional capacity. Prices soared, outpacing the additional production created within the United States. The intended effect of shifting consumption by US manufacturers from imported, mostly Canadian steel to US steel did not materialize. Manufacturers simply moved down the list to the next cheapest alternative, finding esoteric supply chains to substitute Canadian steel. This cycle created a boom and bust scenario, harming US steel producers and manufacturers, with the benefits never materializing. Repeating Mistakes Now, we're considering doing this again. Look at the shale industry. After experiencing a boom and bust cycle, prices rose, and they decided against building additional capacity. They learned from past mistakes and chose not to repeat them. Similarly, US steel and aluminum producers are unlikely to respond to tariffs by expanding capacity. They won't rush to build new refining facilities or start extracting iron ore in Alabama and Mississippi again. Market Dynamics US steel and aluminum producers have found a different market than imported Canadian steel. They have a comparative advantage in quality and time to market. An entire industry exists to provide overpriced cut steel pipe quickly. They charge high prices but deliver swiftly, serving niche markets that require specific products urgently. These producers won't abandon their market to shift to low-end, low-margin steel production. Instead, they'll charge more for the same services, maintaining their market niche. Policy Implications If policymakers would read the case studies, they'd see the flaws in their approach. This isn't a team setting goals and finding experts to implement policies effectively. They're experimenting with a multi-trillion-dollar economy and industries, relying on gut feelings and untested solutions. This approach is unlikely to enhance US competitiveness or yield material benefits. National Security Considerations There's an argument for increasing US domestic steel production for national security reasons. However, this argument is flimsy. To achieve this, investment vehicles and programs are needed to de-risk production expansion. It's similar to what was done with the CHIPS Act, which de-risked chip production in the US, incentivizing companies to take risks before demand was fully reliable. The US government showed a long-term commitment to purchasing US-made goods, even at higher prices. Strategic Partnerships For national security, it's beneficial to have Canada, our largest trading partner and ally, export all their steel, aluminum, and oil to the US. The US industrial policy over the last five decades has tailored supply chains to serve the US market. Forcing Canada to find alternatives would be counterproductive. We want to maintain strong trade ties, ensuring that Canadian industries continue to supply the US, enhancing both economic and national security.
Transcribe →
Tariffs and inflation
Jan 6, 2025Tap to summarizeRant on Tariffs and Inflation Hello, welcome to my thoughts on tariffs and inflation. Here's a bit of background: I sell a digital product online and have a significant number of Canadian and European customers, along with others around the world. My business extends well beyond the United States. The Impact of Tariffs When the idea of tariffs was introduced, I immediately started researching. I realized I needed to read the USMCA and examine Canada's list of retaliatory tariffs on U.S. goods. I had to consider the implications if the U.S. imposed tariffs on Canada, Mexico, or the EU, and if retaliatory tariffs would affect my product. Digital Goods and Tariffs The good news is that the USMCA prohibits tariffs on digital goods. Digital goods are inherently difficult to tariff because they don't cross borders in the traditional sense or go through customs. It's challenging to imagine a scenario where digital goods are specifically targeted without a prolonged trade war or a significant effort from one side. Currently, neither the U.S. nor Canada has the mechanisms to impose tariffs on digital goods effectively. The Real Issue: Customer Confidence The real concern isn't the potential 25% tariff itself—I'd be willing to offer a 25% discount to my customers to offset it. The core issue is the impact on customer confidence in the EU and Canada when considering purchases from U.S. companies. If the U.S. starts imposing tariffs broadly, it could erode trust and confidence in U.S. businesses. Marketing and Messaging To address this, I need to develop a marketing campaign that communicates our stance against tariffs. We could emphasize that we absorb the tariff costs and offer discounts to our customers who are affected by them. Broader Implications Even if digital goods aren't directly targeted by tariffs, U.S. suppliers in Canada and the EU might face negative repercussions. Companies may instruct their procurement teams to be cautious about buying from U.S. companies if alternatives exist, due to the uncertainty tariffs introduce. Inflation: A Misunderstood Term Many claim that tariffs are inflationary. However, the term "inflation" is often misused. Inflation should refer to what we now call "monetary inflation," a purely monetary construct. It's important to distinguish this from price changes, as inflation is the result of market dynamics. In conclusion, while digital goods might be safe from direct tariffs, the broader implications on customer confidence and international trade relationships are significant concerns that need addressing.
Transcribe →
Streaming Apps get zero credit for success
Jan 6, 2025Tap to summarizeThe Challenges of Releasing a Music Streaming App I've been working on a new app, a Progressive Web App (PWA) for music streaming. Despite its incredible performance and how much I enjoy using it daily, I hesitate to release it. Why? It's due to a counterintuitive principle regarding how users perceive utility apps. User Perception of Utility Apps From my experience at HBO, I've learned that certain types of apps are expected to function flawlessly, yet users don't necessarily care about the app itself. These apps often serve as conduits to other content, causing the app to fade into the background. Unlike video games or entertainment media, users don't love the app; they love what it delivers. Consequently, they rarely leave reviews or provide feedback. In essence, you receive zero points for a perfect experience. The Negative-Sum Game of App Reviews The app review landscape is a negative-sum game. On the App Store, I built a tool to aggregate reviews for our app at HBO and analyzed competitor apps. A common pattern emerged: users don't flood the platform with five-star reviews for a seamless experience. Instead, they leave one-star reviews for trivial negative experiences. These are users who have paid for the app for years, use it frequently, and generally have a great experience. Yet, a minor issue can provoke them to leave scathing reviews. For instance, a user might complain about a subtitle color being slightly off-white, despite the app's overall success. As a developer, it can be baffling to see such minor issues result in negative reviews, especially considering the challenges overcome to make the app work. The Reality of Utility Apps When building a utility app or one that serves as a content delivery conduit, this is the reality you face. Users expect perfection because they've paid for it. You get zero points for excelling 29 days of the month; a single negative experience can render your app worthless in their eyes. The phrase "I'd give this app zero stars if I could" is a common refrain in reviews, often for seemingly insignificant problems. These issues, while minor to most, are crucial to a small subset of users. Even if you assure them that a fix is underway, satisfaction isn't guaranteed. The app review scale feels like it ranges from zero to negative five. Conclusion This is why, despite building a music streaming app that I might use for years, I hesitate to release it widely. It may remain a personal project shared with only a few dozen people. I thought I'd share this insight into the challenges of releasing a utility app.
Transcribe →