​

​

​Podcast, YouTube, Blog​

Dear founder,

​

As a software entrepreneur wearing many hats - from coding to marketing to customer service - I’ve found AI systems to be game-changing tools in my daily work.

Today, I want to share how I use these systems, particularly large language models (LLMs) like ChatGPT and Claude, to —hopefully— inspire you to benefit from similar approaches.

This isn’t just about the tools themselves, but about how they’ve transformed my workflow and expanded my capabilities as a solo founder.

​

🎧 Listen to this on my podcast.​

​

The AI Revolution in My Workflow

My approach to work today is significantly different from just a year or two ago. Let me paint you a picture of the “before” and “after”:

Before AI: The Traditional Solo Founder Grind

Two years ago, my work as a software developer and SaaS founder was largely AI-free. Sure, I used tools like Grammarly for writing and some transcription software here and there, but most of my daily grind was just that - a grind. Every line of code, every marketing email, every customer support response - it all came directly from my own fingers and brain.

As a self-proclaimed “1x developer” (and let’s be honest, probably not the greatest at customer service or marketing either), I was stretched thin. I was doing everything myself, from deep diving into database issues to writing product copy, often feeling like I was barely keeping my head above water.

After AI: A New Era of Productivity

Fast forward to today, and the landscape of my work has changed dramatically. AI permeates every aspect of my business, Podscan. Not only is the product itself heavily based on AI for data extraction and analysis, but I’m also using AI tools throughout my business operations.

The key shift has been in how I view these AI systems. They’re not just tools; they’re partners. I don’t delegate tasks to them like I would to an employee. Instead, I collaborate with them, using them as consultants, guides, and even co-founders. They’re there to help me do the work better, not to do the work for me.

Types of AI Tools I Use

Let me break down the main categories of AI tools that have become integral to my workflow:

1. Large Language Models in Chat Contexts: This includes platforms like ChatGPT and Claude. These are my go-to for brainstorming, writing, problem-solving, and even some coding tasks.
2. Specialized Coding Assistance Tools: These are integrated directly into my IDE (I use PHP Storm). They’re always there, ready to help me write better code faster.
3. Task-Specific AI Tooling: This category includes things like advanced transcription tools, which are crucial for Podscan’s podcast processing.

Now, let’s dive into how I actually use these tools in my day-to-day work.

Use Cases and Examples

1. Brainstorming and Ideation

One of the most impactful ways I use AI is for brainstorming and ideation. Let me walk you through a typical scenario:

Say I’m writing a blog post about a new feature in Podscan. Instead of staring at a blank page, here’s what I do:

1. I open up a conversation with Claude or ChatGPT.
2. I give it a clear instruction: “Give me an outline for a blog post about [specific Podscan feature]. Focus on points that would resonate with podcast creators looking for this kind of solution.”
3. The AI provides a bullet-point outline, giving me a baseline skeleton of what most people would want to know.
4. From there, I start filling in my own thoughts and priorities. The AI doesn’t know the intricacies of Podscan or my unique insights as the founder, so this is where I add the special sauce.

But it doesn’t stop there. This is a back-and-forth process. After I’ve written a draft, I might go back to the AI and ask:

• “What might I be missing in this article?”
• “Is there a part of this post that doesn’t quite fit?”
• “Can you spot any arguments that could be easily defeated?”
• “How could I make this blog post stronger and more convincing?”
• “If you were a skeptical reader, what counterarguments would you have?”

This collaborative process helps me create more robust, well-rounded content. It’s like having a writing buddy who’s always available, never gets tired, and has access to a vast amount of general knowledge.

The key here is that I’m not using the AI to write the post for me. I’m using it as a sounding board, a brainstorming partner, and a devil’s advocate. The final product is still distinctly mine, but it’s been refined and improved through this AI-assisted process.

2. Code Generation and Debugging

Now, let’s talk about how AI has revolutionized my coding process. This is where things get really interesting, especially for my fellow developers out there.

The Old Way

In the past, when I needed to implement a new feature, my process looked something like this:

1. Think about the feature and how to implement it.
2. Start writing code, frequently pausing to look up documentation.
3. Run into issues, debug, rinse and repeat.
4. Finally get a working version, then refactor and optimize.

This process could take hours, even for relatively simple features.

The New AI-Assisted Way

Now, my process looks radically different. Let me walk you through a real example:

Let’s say I need to implement a function to calculate the average word count of the last 10 podcast episodes transcribed on Podscan. Here’s what I do:

1. I open my PHP file in PHP Storm.
2. I create a new function with a descriptive name.
3. Inside the function, I write a comment that describes what I want the function to do, like "Load the latest 10 episodes and summarize the number of words in each of its transcripts "
4. I hit Enter after the comment, and within seconds, my coding assistant (powered by AI) generates a solution, in under a second.

​

Now, here’s the kicker: In about 80% of cases, this generated code is exactly what I need or very close to it. It understands the context of my application (it knows what an Episode is and how to query it), and it implements a straightforward solution to the problem.

But my job isn’t done. I still need to review this code, test it, and potentially refine it. Maybe I want to add error handling, or perhaps I need to account for episodes without transcripts. The AI has given me a huge head start, but I’m still the one steering the ship.

This approach has dramatically changed how I think about coding. Instead of focusing on writing the code as quickly as possible, I now focus on writing the most precise and clear comment (or prompt) I can. The better my prompt, the better the AI’s output, and the less work I have to do to refine it.

It’s teaching me to be clearer in my specifications, which is a valuable skill in itself. And the time savings are enormous - what might have taken me an hour now often takes just a few minutes.

3. System Architecture: A Real-World Example

Now, let’s dive into a more complex example that really showcases the power of AI in software development. Recently, I completely reworked what is probably the most high-throughput, central system of Podscan: its internal podcast queuing system.

The Problem

The existing queue system was a resource drain. It was sitting on top of a database, causing a lot of database queries to figure out the next item to process, count items, and manage priorities. With thousands of items in the queue, this was putting strain on my database for other requests and API calls.

The AI-Assisted Solution

Instead of diving straight into coding, I started a conversation with Claude AI. Here’s roughly how that conversation went:

Me: “I want to build a queuing system for my podcast transcription based on Redis, using the Redis commands that are highly performant from inside PHP. Here are the requirements:

• It needs to have at least four different kinds of priority
• It needs to grab new items for the highest priority first, and the lowest priority last
• I need to be able to add new items on the fly
• The collections in which these queue items exist need to be sorted by date, so I can always pick the newest episode first
• I need metrics: queue size, processing time per item, estimated time to process the whole queue
• This needs to be a standalone service in a Laravel app

Can you help me design this system?”

Claude then started generating code, and because I gave it such a holistic perspective on what the service was supposed to do, it created an initial version that I could almost directly copy and paste into my application.

This was just the starting point. As we continued our conversation, I realized I needed a few more things:

1. A way to preload existing items from the database into the queue
2. A method to estimate processing time based on historical data

Claude quickly added these features.

The result? A high-performance system capable of handling millions of podcast episodes. It’s mind-blowing to me that this critical part of my application was entirely written by an AI system, with me primarily doing code review and providing guidance.

What’s even more impressive is the scalability. Claude helped me do some quick calculations, and we found that even if I were to queue every single podcast episode that ever existed (we’re talking about 170 million episodes), the system could handle it without breaking a sweat. The memory footprint would still be manageable on my existing application server, using less than 40% of the available RAM.

This is a massive improvement over the previous system, which would slow to a crawl with just 20,000 items. Now, I can finally import every single podcast episode out there that I can find in the feeds I read every day - we’re talking millions of episodes daily.

The impact on Podscan’s capabilities is enormous, and it all started with a conversation with an AI.

4. Debugging Server Issues: AI as a Systems Admin

Now, let’s switch gears and talk about how AI has become my go-to assistant for troubleshooting server issues. As a solo founder, I wear many hats, and sometimes those hats don’t fit quite right. Systems administration is definitely one of those for me.

The SSH Connection Problem

Recently, I ran into an issue where I couldn’t connect to one of my backend servers. This was preventing me from deploying updates, which, as you can imagine, is a pretty big problem. Here’s how I tackled it with the help of AI:

1. I took the error message from my deployment system and my failed login attempts.
2. “I have a Ubuntu server that I’m having trouble connecting to. Here are the errors I’m seeing: [pasted error messages]. What could be causing this? How can I diagnose the problem further, and what are some potential solutions?”
3. Claude analyzed the error messages and suggested a series of steps to diagnose the problem. It explained each step and why it might be relevant.
4. Following Claude’s advice, I checked various log files on the server (once I managed to get in through a different method). I copied relevant lines from these logs and pasted them back to Claude for further analysis.
5. Through this back-and-forth, we (yes, it felt like a collaborative ‘we’) identified that the issue was related to a configuration in the SSHD config file. Claude suggested a specific change to make.
6. Before making the change, I asked Claude to explain the implications of this change and any potential security considerations. Once I was satisfied with the explanation, I made the change, and voila! I could connect again.

But Claude didn’t stop there. It also suggested I look into reactivating Fail2Ban, a tool that helps protect against brute force attacks. It walked me through the process of setting that up as well.

The Redis Connection Mystery

In another instance, I was having trouble connecting to my Redis instance. This time, the process went something like this:

1. I shared the error messages with Claude and asked for potential causes.
2. Claude suggested it might be related to a recent system update and advised me to check specific log files.
3. I found the relevant logs, and sure enough, there had been an automatic software update right before the problems started.
4. Claude helped me interpret the log entries and suggested a rollback of the problematic update.
5. After successfully rolling back, Claude guided me through the process of adjusting my automatic update settings to prevent similar issues in the future.

What amazes me about these interactions is not just that Claude could help me solve the problems, but how it walked me through the process, explaining each step. It’s like having a patient, knowledgeable systems administrator available 24/7, willing to explain things in detail and never getting frustrated with my lack of expertise.

5. Documentation and Standard Operating Procedures (SOPs)

One of the challenges of being a solo founder is keeping track of all the processes and procedures you develop over time. After resolving issues like the ones I just described, I’ve made it a habit to create documentation or Standard Operating Procedures (SOPs) for future reference. And guess what? AI helps me with this too.

Here’s my process:

1. I draft the document, laying out the steps I took to resolve the issue.
2. “I’ve written an SOP for handling SSH connection issues. Can you review it and let me know:
   • What might be missing?
   • What might a future employee (or future me) misunderstand?
   • Is there any erroneous or sensitive information that shouldn’t be included?
   • How can I make this more clear or comprehensive?”
3. Claude then provides feedback, often catching things I’ve missed or suggesting clearer ways to explain certain steps.
4. I revise the document based on this feedback.
5. Finally, I might ask Claude to generate a quick summary or checklist based on the SOP, which I can use as a quick reference in the future.

This process helps me create more robust, clear, and comprehensive documentation. It’s like having a fresh set of eyes look over my work, catching potential issues before they become problems down the line.

6. Customer Communication: AI as a Communication Coach

As a solo founder, a significant part of my day involves communicating with customers. Whether it’s responding to support tickets, writing marketing emails, or crafting product updates, clear and effective communication is crucial. Here’s how AI has become my indispensable communication coach:

Drafting Responses

When I receive a customer inquiry or support ticket, my process typically looks like this:

1. I draft an initial response based on my understanding of the issue and the solution.
2. “I’ve drafted this response to a customer who’s having trouble with [specific issue]. Can you review it and suggest improvements? Consider tone, clarity, and completeness. Also, is there anything I might be missing or any follow-up questions I should ask?”
3. The AI then provides suggestions for improvement. It might point out areas where I could be clearer, suggest a more empathetic tone, or remind me of additional information I should include.
4. I revise my response based on this feedback, making sure to maintain my own voice and style.

Using AI-Powered Customer Service Tools

I also use tools like Crisp Chat, which has a “magic reply” feature powered by AI. Here’s how I incorporate this into my workflow:

1. When a new customer message comes in, I first look at the AI-suggested response.
2. This gives me a quick starting point and often reminds me of key points I should address.
3. However, I never send these suggested responses verbatim. Instead, I use them as a base, heavily modifying them to match my tone and to include specific details about Podscan.
4. If I’m unsure about the modified response, I might run it by Claude for a final check.

This process helps me respond more quickly and consistently, while still ensuring that each response is personalized and genuinely helpful.

Crafting Marketing Emails

For marketing emails, my AI-assisted process looks something like this:

1. I start by outlining the key points I want to cover in the email.
2. I then ask the AI to generate a draft based on these points, specifying the tone I’m aiming for (e.g., professional, friendly, excited).
3. Once I have the AI-generated draft, I heavily edit it, infusing it with my voice and adding specific details about Podscan that the AI wouldn’t know.
4. Finally, I might ask the AI to review my edited version, looking for ways to make it more engaging or persuasive.

This collaborative process helps me create marketing emails that are well-structured and engaging, but still authentically mine.

The key in all these communication tasks is that I’m using AI as a tool to enhance my own abilities, not as a replacement for my own judgment and personal touch. It’s like having a really smart friend who’s always available to brainstorm and provide feedback, but who doesn’t know the intricacies of my business or my relationships with my customers.

Why I Spend So Much Time with AI Tools

You might be wondering, “Arvid, this sounds like a lot of back-and-forth with AI. Is it really worth the time?” Let me break down why I find it so valuable:

1. Cost-effectiveness: At $20-30 per month for most AI services, it’s incredibly cheap compared to hiring human assistance. As a bootstrapped founder, this is a game-changer.
2. Always available: Unlike human employees or contractors, these AI tools are there 24/7. Whether I’m working late at night or early in the morning, I always have a capable assistant at my fingertips.
3. Versatility: These tools help me become better at everything I do. They make me a better marketer, a more efficient coder, and a more responsive customer service rep. For a solo founder wearing many hats, this versatility is invaluable.
4. Time-saving: While it might seem like I’m spending a lot of time interacting with AI, it’s actually saving me time in the long run. The AI handles about 80% of the initial work, leaving me more time for strategic thinking and refinement.
5. Knowledge gap filler: There’s so much I don’t know, especially in areas outside my core expertise. AI provides immediate answers to questions I might not know, like specific server configurations or log file locations. It’s like having a team of experts on call.
6. Consistency: Using AI helps me maintain a consistent level of quality across all aspects of my work, even in areas where I’m less skilled.
7. Learning tool: Interacting with AI often teaches me new things. Whether it’s a coding technique I hadn’t considered or a marketing strategy I wasn’t aware of, I’m constantly learning.

Limitations and Challenges

While AI has become an integral part of my workflow, it’s not without its challenges. Here are some limitations I’ve encountered and how I deal with them:

1. Accuracy and reliability: While rare, AI can provide incorrect information or suggest problematic solutions. This is particularly critical for tasks like server management or security-related issues. To mitigate this, I always verify critical information, especially for server commands or security tasks. I treat AI suggestions as a starting point, not gospel.
2. Need for oversight: I don’t fully delegate tasks to AI. I remain involved in the process, reviewing and verifying outputs. This means I still need to understand the underlying principles of what I’m doing, whether it’s coding, marketing, or system administration.
3. Potential over-reliance: There’s a risk of becoming too dependent on AI tools. I’m cautious about this, ensuring I still understand how the work is done and can do it manually if needed. It’s important to use AI as a tool to enhance your abilities, not as a crutch.
4. Keeping up with changes: AI tools are evolving rapidly. What was cutting-edge a few months ago might be outdated now. I need to stay aware of new capabilities and limitations, which requires ongoing learning and adaptation.
5. Context limitations: While incredibly knowledgeable, AI doesn’t know the specifics of my business or customers. I always need to add that context myself and verify that the AI’s suggestions make sense for my specific situation.
6. Ethical considerations: As AI becomes more integrated into my work, I have to be mindful of ethical considerations, particularly around data privacy and disclosure. I’m always transparent with my customers about how and where AI is used in my processes.

Best Practices I’ve Developed

Through my journey of integrating AI into my workflow, I’ve developed several best practices that help me get the most out of these tools:

1. Understand prompting: The quality of AI output is highly dependent on the quality of your input. For creative tasks, I’ve learned to ask for variety in results. For specific tasks, I aim to be as detailed as possible in my prompt. It’s a skill that improves with practice.
2. Treat AI like a co-worker: I’ve found it helpful to approach AI interactions as I would conversations with a very knowledgeable colleague. I learn how to communicate effectively with the system, refining my approach based on the results I get.
3. Never use results verbatim (except sometimes with code): Whether it’s a written response or a suggested solution, I always review, test, and verify AI outputs. Even with code, while I might use AI-generated code directly more often, I always review and test it thoroughly.
4. Follow a process: I’ve developed a general workflow: reflection before prompting, then prompting, then verification of results. This helps ensure I’m using AI effectively and not just throwing questions at it haphazardly.
5. Use AI for iteration: I often use AI in multiple rounds for a single task. For example, I might generate an initial draft, review and modify it, then ask the AI for feedback on my modifications.
6. Combine AI with human knowledge: I’ve found the best results come from combining AI capabilities with my own expertise and understanding of my business.
7. Stay curious: I regularly experiment with new ways to use AI in my workflow. Sometimes these experiments don’t pan out, but often they lead to new, more efficient processes.

The Future of AI in Software Development

As I use these AI tools day in and day out, I can’t help but get excited about the future. Here are some developments I’m looking forward to or anticipating:

1. Specialized models: I’m excited about the potential for more specialized models trained on specific codebases. Imagine an AI that understands not just your current code, but the full history of your project, including experiments that didn’t work out. This could be incredibly powerful for maintaining and evolving complex systems.
2. AI-driven product planning: We might see AI systems that can analyze code, customer conversations, and social media mentions to suggest the next steps for product development. This could help solo founders like me make more informed decisions about where to focus our limited resources.
3. “AI Ops” specialists: I can see a new role emerging - something like an “AI Ops” specialist who orchestrates AI systems within an organization, ensuring they have full context and are properly integrated. For larger companies, this could be a crucial role in maximizing the benefits of AI.
4. AI in software education: I believe AI will become an integral part of software development education. We’re already seeing this with tools like GitHub Copilot, but I expect it to go much further. We might even see programming languages shipping with built-in language models to facilitate development.
5. More integrated AI tools: While I currently use a combination of standalone AI chat interfaces and IDE-integrated tools, I expect we’ll see more comprehensive, deeply integrated AI assistants that understand your entire development ecosystem.
6. AI-assisted code review: Beyond just generating code, I’m excited about the potential for AI to assist in code review, helping catch potential bugs or suggesting optimizations.
7. Natural language programming: As AI gets better at understanding and generating code, we might move towards a future where more programming is done through natural language instructions, with AI handling the translation to actual code.

Solo But Not Alone

For solo entrepreneurs and founders, I can’t recommend integrating AI tools into your workflow strongly enough. They can significantly impact your productivity and capabilities, allowing you to punch above your weight in a competitive business landscape.

Look for opportunities to use AI in any task involving writing or problem-solving. Whether it’s coding, marketing, customer service, or strategic planning, there’s likely a way AI can help you do it better.

Using AI assistants is the new form of early hiring - getting consultants that cost you $20 a month in aggregate for every aspect of your business. It’s incredibly empowering, and I’m excited to see how these tools will continue to evolve and support entrepreneurs like us.

Remember, the goal isn’t to replace your own skills and judgment, but to enhance them. Use AI as a collaborator, a brainstorming partner, and a knowledge extender. With this approach, you can achieve things that might have seemed impossible for a solo founder just a few years ago.

As we move forward in this AI-enabled world, stay curious, keep experimenting, and don’t be afraid to let these powerful tools amplify your capabilities. The future of entrepreneurship is here, and it’s augmented by AI.

​

​

If you want to track your brand mentions on podcasts, please check out podscan.fm — and tell your friends!

​

​

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!

​