​

​

​Podcast, YouTube, Blog​

Dear founder,

​

As I watched the latest Apple Keynote from WWDC, one sentence really stood out to me: "AI for the rest of us."

Apple is going to introduce artificial intelligence into their devices, software, operating systems, and apps in a way that will make it ubiquitous within a few months, maybe even weeks, once people start downloading and installing the latest beta version of iOS.

The question is, how will that impact consumer expectations around AI?

​🎙️ Enjoy the blog post and the podcast

​

​

Will it be for better or worse when it comes to us as entrepreneurs and the products we're building? Today, I'm going to explore the current state of consumer-facing AI and how a major player introducing it into their hardware and software might influence that, and how this affects solopreneur and bootstrapped software businesses.

This episode is sponsored by acquire.com. I am Arvid Kahl, and this is the Bootstrapped Founder podcast.

The first big expectation will be that people are going to get used to the quirks of chat-based interfaces and their limitations. Most people using ChatGPT now came to it via recommendation from someone who also used ChatGPT. It started in the technical community and made its way into marketing, HR, and creative departments. But it always came through a chain of recommendations from technical people who understood the features, complexity, and limitations of the system. Obviously, someone who does not understand these limitations will have heightened expectations, but there was always a way to set foundational basic expectations, and for ChatGPT, over its development cycle, to slowly become better at them.

I think this will be significantly different once Apple introduces free ChatGPT usage to every single iPhone out there through their latest iOS update. Millions, if not billions, of people will, for the first time, be exposed to the shakiness, potential instability, and imperfections of AI systems and their responses, no matter how good Apple is at implementing their own local LLM on their devices or integrating with ChatGPT and GPT-4 or whatever successor may come in the next few months.

At this scale, consumers will reliably experience that AI systems are still gaslighting engines. They confidently talk about things they don't necessarily understand and will try to convince you that their answer is correct and trustworthy. This has a couple of consequences. First, it might cause significant damage on occasion to individuals who will then stand out in their opposition to AI features. Second, it will also train us that these models, even though they occasionally don't give a good response, will always give some response for us to continue working on. We will have tools that, like an unsafe weapon, can go off on their own, and also tools that we understand are fallible, limited, and require a second opinion.

As these tools continue to amaze people who use them correctly and confuse those who use them incorrectly, we will see a learning experience on a social level. Depending on your generation and how open you are to new technologies, people will learn that AI tools are really useful under supervision. Over the next few years, the sentiment already clearly established in the technical community will spill into the consumer community. Right now, people often get quickly disillusioned by the limitations of AI. But once AI is integrated into a system where it interacts with your calendar, email, and voice messages, becoming a transmitter of sorts, a transitionary processor of information that makes it more dense, actionable, and easier to deal with, people will see AI as a necessary tool to facilitate and speed up interactions they would generally want to have anyway. They will move from seeing ChatGPT as AI to seeing AI as a holistically integrated tool, a part of the operating system.

​

And since we're talking about a holistic approach to building things: let me introduce the sponsor of this edition, Mixo!

​

Just like in the Windows and Mac worlds, where Quick Search and tools like Alfred or Raycast are integrated into software stacks as a means to quickly get what you want, we will see AI tools becoming that on an operating system level. People will not go to ChatGPT or Anthropic or wherever to do something anymore. They will have a standardized channel to ask AI agents to either do something for them or fetch information that helps them do the thing themselves. For AI to be truly effective, it must work seamlessly across different platforms and devices, requiring standardizing protocols and ensuring compatibility.

One of the bigger and more important developments in AI is that we will likely see a living experiment with Apple connecting you to GPT and their own language models, which will receive updates every few months. The data Apple collects on the device, if you allow them to, will train better models, or they might even build an infrastructure on the device or in your network to train local language models for your own usage. We might see training for specific personal purposes on the device that inference will happen on. Over time, this cycle between training and inferencing will be expected for every kind of software that uses AI. It's logical that if you have a text processor like Grammarly, which already uses AI features, it will likely use an AI model trained exclusively on your writing style. You might even use writing style models of other people if they choose to open these models. The implementation of AI should be accompanied by robust regulatory frameworks that set standards for safety, privacy, and accountability.

If you use AI as a facilitator in your application, either as an assistant or a means for people to communicate with each other or with your system itself, future expectations will probably include a personal approach for that assistant or communicator to work with you or in your stead. It will be trained on unique data to work on unique data. This is a long-term expectation, but you might want to think about it from the start. What data do you need to collect? How do you need to protect it to facilitate training very personal, edge-specific models for your customers? Ethical AI is critical. Developers must ensure that AI systems are designed and trained to be fair, transparent, and unbiased.

There will also be a perception of being able to initiate almost anything by writing it as a human language-centric instruction. With iOS and macOS versions of these language models being integrated into the system, we will move from step-by-step, point-and-click instructions to very simple instructions. For example, in the Apple calendar app, you can enter a sentence, and it will find the date, purpose, and alarm settings from that sentence. People will expect this in almost every piece of software. They will go into Adobe Photoshop and ask for an image 800x600, half black, half white, with text in the middle, and a black and white photo of an old Baroque mansion as the background. They will go into Microsoft Word and say, "Write an outline for an article on the Galapagos Islands, make it five pages long, and include several pictures sourced from Wikipedia." They will go into their VS Code or PHP Storm or Vim and say, "Create the scaffolding for a REST API that has API key support, is built on Laravel, and allows people to upload PDF documents to S3."

These complex initial commands will likely become more common, and so will people's expectations of how complex these instructions can be. This doesn't mean you have to implement it in your software, but it's interesting to consider how you can allow your customer to prompt your software to do things for them instead of clicking through your UI. There are WebGL or WebGPU tools that allow for almost instantaneous transcription between the microphone and your server, and these tools will soon become common libraries integrated into almost every front-end library. If this is the case, what could you facilitate for your customers? What could you give them that does something in seconds that would take them minutes? The environmental impact of AI, particularly in terms of energy consumption for training large models, is a concern that needs addressing.

AI as a smart tool in software businesses built by solopreneurs and bootstrapped founders is manageable in the browser. You could have all the actual inference happening in the browser itself, provided people download a small model, maybe 50 or 200 megabytes in size. If Apple brings their AI to the edge device, you will eventually have access to that model through system APIs. Browsers will likely implement this soon. It's likely that Chrome, Firefox, Safari, and others will have a native JavaScript API to access small but reliable local language models. It might require a few hundred megabytes of downloads, maybe a gigabyte, to have a small version of Llama 3 or whatever Apple or Google develop next. The development of small language models is fast-paced, and we will see better builds within the next few months. If the browser or your mobile system has immediate access to this technology, at least for prompting your tool to do an action for your customer, that might be extremely useful. Any search feature you have could be extended to allow for AI-based searches on the data in your customer database through prompts provided by your customers.

Ultimately, we will have to deal with AI expectations from normal users who have been exposed to this technology on their mobile devices and will expect it in their professional or consumer software sooner rather than later. Think about what you can do to facilitate training models on the edge for specific tasks your users might have. Consider how you can enable your customers to prompt your software to do things for them instead of clicking through the UI. And think about what you can do in the background to facilitate processes using AI technology that were previously based on regular, non-AI augmented technology. Building public trust in AI is essential for its widespread adoption. Transparency in AI processes and clear communication about the benefits and limitations of AI will be key to gaining user acceptance.

​

I'll share a few updates about my SaaS on the pod, and if you want to track your brand mentions on podcasts, check out podscan.fm!

​

Classifieds

I recently launched The Bootstrapper's Bundle, which contains Zero to Sold, The Embedded Entrepreneur, and Find your Following. If you want to start a bootstrapped business and build a validated product and a personal platform while doing it, check out this bundle. It contains all eBooks, audiobooks, video courses and extra materials I ever created. It's just $50, for now.

​

​

Thank you for reading this week’s essay edition of The Bootstrapped Founder. Did you enjoy it? If so, please spread the word and ​share this issue on Twitter.​

If you want to reach tens of thousands of creators, makers, and dreamers, you can ​apply to sponsor ​an episode of this newsletter. Or just reply to this email!

​