207: ChatGPT: Goldmine or Minefield for Indie Hackers?

Download MP3

Hello everyone and welcome to The Bootstrapped Founder.

My name is Arvid Kahl and I talk about bootstrapping, entrepreneurship and building in public.

Today I'll talk about what bootstrappers can expect from a future full of systems like


We'll talk about the good, the bad and the unexpected.

The wave of AI tools is just building up somewhere in the very very blue ocean.

We better get ready for the tsunami that will hit us once all of this becomes a mainstream

part of building a business.

And that brings me to the sponsor of today's episode.

I've got another question for you.

Why did the business owner open a store at the bottom of the ocean?

They wanted to liquidate their assets.

Speaking of assets, do you have enough operating cash to make it through to the next month?

Enter Pinto Financial.

They will streamline your bookkeeping, your forecasting and cash flow but they don't stop


Pinto takes it to the next level, providing custom, video-based, monthly reports so you

always know where you stand.

Think of them as your financial lifeline, guiding you through the choppy waters of business

financial management.

So if you're ready to navigate these waters with confidence, visit usepinto.com to learn

more about how Pinto Financial can help you build profitably.

Don't let your cash flow drift away.

Choose Pinto Financial and let's make waves together.

Alright, let's get back to our main topic today.

Everybody in the indie founder community is talking about chat GPT.

Some founders discover how powerful it is as a writing tool and others have it built

their products and then some entrepreneurs are even building businesses on top of open

AI's conversational, chat-based AI.

Beyond that, it's also a meme now.

Everyone is talking about how everyone is talking about it.

It's taking the business world by storm and where there is change, there is also opportunity.

And risk, potentially massive risk.

So a warning here, just like most of the things that chat GPT hallucinates, none of what I

will talk about here is the factual, absolute truth.

It's future talk, a prediction at best.

I've seen a few things come and go and I believe that what follows is a likely scenario,

but it is a prognosis nonetheless.

And another thing, while I talk about chat GPT specifically here, we'll see new versions

and variations show up over time.

I'm referring to the whole lot, all of them, not just what we get to work with right now.

So let's gaze into the crystal ball.

Here's what a future with chat GPT will very likely mean for you as an indie founder, your

business, and the reality of your customers.

The first thing, congratulations, you just found a co-founder that never sleeps.

No matter how much you involve chat GPT in your life, it will always be available.

There downtime, windows excluded, but conversational AI will soon be as ubiquitous as the internet

already is today.

And while it is already a two-sided chat, it will soon become much more than that.

Once chat GPT becomes an agent that can interact with the live internet through reading or

posting to websites, making calls or moving files around, you will have a digital demon

ready to do your bidding, anything it could possibly do for you within milliseconds of

you uttering your wishes immediately.

As open AI rolls out their plug-in ecosystem, more and more chat GPT powered tools will

remove obstacles from our days, from personal scheduling through negotiating and booking

customer interviews to helping you solve business strategy problems, you'll likely always check

in with chat GPT first.

The interesting challenge will be to learn where and when chat GPT cannot be relied upon.

The conversational AI is effectively gaslighting you, it's a gaslighting machine, it's trained

to create believable and convincingly detailed responses to any questions you might ask,

which means that you will need to be careful with anything the system tells you to do.

At some point, you'll need to involve an actual human being to ensure you're not just blindly

following the whisperings of the ghost in the machine, and that will be the defining

competency of a whole generation, judging the commands and the results of an interaction

with conversational AI systems.

It's like how we grew up to be good on the internet to understand things natively that

will happen with AI too.

It's a skill, just like using Google.

Using chat GPT is just another skill, so don't miss out on it.

Those who ignored search engine optimization now struggle to keep up with being visible

on the internet.

Even if you learn just the basics of SEO, you'll quickly outrank those who don't.

Conversational AI, and AI in general, will be just the same.

Understand that it is a tool that facilitates the transformation of data.

It's an always-on assistant that's effectively free and growing more powerful daily.

Why would you not learn to use such a powerful tool?

One thing is obvious though.

Things will change incredibly fast.

They already do.

Barely four months have passed between chat GPT's release in November 2022 and the arrival

of the massively improved GPT-4 in March 23.

The interfaces are changing, the prompts we use to communicate are shifting around, and

new mind-bending use cases are discovered every single day.

It's almost impossible to keep up with the best practices as they evolve daily too.

I personally 80-20 this.

I check in regularly with the newest developments, but I don't waste time on just staying on

the edge.

I have more pressing things to do, and I recommend not over-investing your time into this either.

Because the speed at which the landscape changes is a problem.

We cannot build long-term businesses on constantly fluctuating technology.

Sure, we can exploit this massive surge of interest right now, and some people do, but

nobody enjoys building a business that lasts a few months before another quantum leap makes

it obsolete.

So is all lost?

Should we just ignore this business opportunity as bootstrappers?

Well, let's see.

There is one massive benefit to chat GPT.

It is a platform that drastically lowers the barrier to entry and empowers non-technical

founders to build incredibly complex software prototypes.

But there is also an extremely problematic drawback to chat GPT.

It is a platform that drastically lowers the barrier to entry and empowers non-technical

founders to build incredibly complex software prototypes.

Same deal.

Let's dive into the dual nature of each point of this argument here.

It lowers the barrier to entry.

Even at its current, early incarnation, chat GPT will happily converse with you about the

steps you need to take to come up with business ideas or validate them in your market of choice

and how to turn them into a sustainable business.

All these steps the machine knows because it has read all the books and articles on

the subject and it can tell you.

It's never been easier to get up to speed on what to do to get started.

The problem here, chat GPT has this very conversation right now with another eager founder to be.

Probably a couple hundred people at the same time right now.

It's becoming easier for everyone.

It also makes software prototypes incredibly easy to build.

Now, no code made prototyping very easy already, but chat GPT might even remove that abstraction.

It will speed up prototyping even more.

An example?

I don't know how to code Swift.

And I know how to code, but Swift on Mac OS, I don't know that.

But I wanted to build a little tool that would automatically back up my SD cards whenever

I plugged them into my Mac.

I asked chat GPT to help me write the code for this tool, and line by line, the AI told

me where to download Xcode, how to create a new project, and then fed me exactly what

code to put into which file.

Fifteen minutes later, I clicked run and I had a working version of what I wanted.

Now, I'm a developer and I grasp these things quickly.

I could swiftly take it from here and add in-app purchases to turn this into a business

or something like that, but even a non-coder can reach that prototype stage within hours.

Real products, not just pitch decks.

Chat GPT is powerful.

So expect competition even earlier than before.

And it's also a platform.

Building on top of chat GPT, using either its API or building plugins for it, is a massive

opportunity for first movers and creates an equally massive platform risk over time.

You'll probably get a lot of attention when you're the first to build a particularly

powerful AI-powered tool, but expect it not to last for too long.

Between competitors eating into your profits and OpenAI being in complete control of just

how much, and if at all, you can use their interfaces, you're constantly exposed to

more risk than if you could run an AI like this on your own infrastructure.

And we'll talk about that later.

But monetize accordingly.

Charge upfront instead of over time, diversify as much as you can, and actively watch open

source implementations to see if you can spin them up yourself as a backup or alternative

should you ever be deplatformed.

This is sobering.

Maybe AI isn't a good fit for bootstrappers after all.

Here's the way out.

Make AI an ingredient, not a dish.

AI as a feature will be the more sensitive long-term perspective for bootstrappers rather

than built on AI.

If using an AI system improves your overall product, it becomes less of a liability and

more of an amplification engine.

Not everything needs to be AI-based either, not all work is smart, and some things benefit

from simplicity and clear structures which can traditionally be accomplished by straightforward

software tools.

Now how would you then evaluate this situation?

What is AI suitable for and when should it be avoided?

The answer is context.

Your product solves a specific problem among the many many challenges that your customers


Every software product is part of a specific workflow.

Sales data goes in, a PDF report comes out, an unedited video file goes in, a ready-to-upload

transcript and a download link to a subtitle file come out.

Your product is a value transformer, and anything that helps it transform the inputs into the

outputs better is a good place to be enhanced by AI.

But if you have to train your customers to use a new feature just because it contains

AI, completely outside their existing workflow too, you're reducing their overall value

perception of your product.

Even if you think this new feature is great and will revolutionize how people work, you

won't likely convince more than a few early adopters to change the way they work.

And that's alright when you're just starting out, it's kind of how most businesses try

to get into the door somewhere, but any sustainable business operates through sustainable practices.

So understanding the value flow around your product's real-world use is way more important

than slapping the AI-powered sticker on the box.

But no matter if you integrate AI into your product or just keep building traditional

software, ChatGPT will be a helpful co-programmer.

Its capacity to write code is just as impressive as its skill at composing compelling headlines,

convincing articles, and even clever jokes.

When you pair up with a conversational AI, all writing will be much easier, which means

you should most definitely write, or start writing if you're not already.

Founders benefit immensely from communicating with their peers and customers through the

written word.

ChatGPT is a powerful catalyst for those of us who generally shy away from writing, because

we're not used to it.

And it has gotten really, really good at writing.

So good that it's already taking the jobs of experienced writers, particularly in the

freelance field.

If a free AI tool can do that, not integrating it into your writing routine feels almost


Sure, you should never use its output verbatim, but it can be an amazing brainstorming tool

early in the process.

And that, I believe, is its actual superpower.

It might not be good at math, and it comes up with things that are entirely untrue, but

the conversational nature of this AI system will make it a widely used companion for all

kinds of activities.

Obviously, copywriters will use it for fresh content ideas, that's what everybody's

doing right now, but you will find a version of ChatGPT that will find its way into doctors'

offices, therapists' couches, and all over the fitness industry.

We will see judicial applications, just as much as there will be salespeople asking ChatGPT

for the perfect pitch for that one person just before they walk up to that person's

front door.

It will be everywhere.

And those things tend to be very hard to contain.

It will leak into everything.

We're already experiencing ChatGPT taking over the world by storm, and not just the

business world.

Over the recent Easter holidays, several of my peers reported chatting about ChatGPT with

their non-technical relatives.

I did a poll on Twitter, it's like 45% of people talked about this over Easter.

And their relatives already knew about it before.

It's seeping into the professional world and the private world as well.

And there's a different kind of leaking happening here too.

The actual technical implementation and the weights of the Llama model, or most simply

the code and configuration behind a research version of conversational AI like ChatGPT,

have been leaked.

And that's spurted a massive avalanche of open source implementations.

And once the cat's out of the bag, people will jump at the chance to build things and

use them for themselves.

We will see this accelerate even further.

Some projects have even been built and released into the open source world voluntarily, like

OpenAI's Whisper.

That's a tool that generates preliminary captions and transcripts for this very show as well.

I use that AI tool for my own content.

So once people have the tools to train their own language models, we will see something

else that's super interesting.

We'll see a Cambrian explosion of AI diversity.

We'll have AI systems exclusively trained on decades of content from one particular

community forum.

Users will expose inherent biases stemming from how we select training data and then

figure out ways to unbiased the AIs of the future.

It's going to be wild.

And this development will be crucial to AI adoption.

We need to move AI systems from centralized services, such as OpenAI server clusters where

ChatGPT and GPT-4 and all these things currently reside, towards the edge of phones, our tablets

and our home computers.

AI needs to be usable offline and it needs to be air gapped to the device that it's being

used on.

Otherwise, privacy regulators will make it unusable, and rightfully so.

We had an incident already.

Samsung employees recently leaked trade secrets by using ChatGPT for their work tasks.

And no business will take this lightly.

Unless we can be absolutely sure that our inputs remain out of the hands of tech giants,

we will see these tools banned swiftly, first in corporations and then through regulation


They are too alluring not to be regulated.

But once we can run a full ChatGPT on our phone, AI will allow for privacy-sensitive

use cases.

Businesses will run an on-premise version where business-critical confidential data

is actively fed into the system to then train models that can automatically suggest business

strategies and onboard new employees and detect inefficiencies, in a way that makes the models

themselves a valuable item, not just the data.

AI model protection will be a job.

I can already imagine this James Bond movie where the villain is trying to steal the model

so that they can be one step ahead of the secret service which uses AI for threat detection

stories right themselves, almost literally at this point.

We'll see models for all kinds of specific use cases pop up, and this will strongly affect

the meaning of expertise as well.

The more AIs are trained on what's already out there, the more general expertise will

be absorbed into the AI systems of the future.

What's left for the human experts are all the non-generalizable things.

You will become even more of a niche subject matter expert if you want to be able to provide

more than ChatGPT and its successors can provide.

Dive into your niche, nerd out without restraint.

If AI systems have to appeal to a general public, well then you must lean into the not-so-general

public of your fellow niche inhabitants.

As you can probably tell, I'm quite optimistic about this.

Other voices, like Les Rungs, Eliezer, Jutkowski, are terrified of the AI systems of the future,

and they have solid arguments to be cautious.

Realistically, we could debate AI's usefulness and ethical implications for weeks, and during

those weeks, researchers will have made great strides forward towards better models, and

clever entrepreneurs will have leveraged that technology to build something meaningful that

others are willing to pay for.

So focus, right?

Don't let yourself be dragged away from the pragmatic nature of these tools right now

because of all the hype, or your staunch dismissal of all things AI.

It's not worth to take a stance here.

If you're annoyed by AI because everyone is talking about it, well you should ask yourself

why this is such an important topic, because not all hype is unfounded, particularly when

it's not just something abstract, the tool is right here.

You can use it, literally, to write a book about anything that you're interested in

within a few hours.

That is significant.

If you ever wanted to give in to your FOMO, this might be the best time to do it.

Learn how to use ChatGPT for yourself, and your business, AI will stick around, and if

you understand it, well, so will you.

And that's it for today.

Thank you to Listening for the Bootstrap Founder, and thank you to Pinta Financial for sponsoring

this episode.

You can find me on Twitter at arvidkal, A-R-V-I-D, K-A-H-L.

You will find my books and my Twitter course there as well.

If you want to support me and the show, please subscribe to my YouTube channel, get the podcast

in your podcast player of choice, and leave a rating and a review by going to ratethispodcast.com

slash founder.

Any of this will really help the show.

Thank you so much for listening, and have a wonderful day.

Bye bye.

Creators and Guests

Arvid Kahl
Arvid Kahl
Empowering founders with kindness. Building in Public. Sold my SaaS FeedbackPanda for life-changing $ in 2019, now sharing my journey & what I learned.
207: ChatGPT: Goldmine or Minefield for Indie Hackers?
Broadcast by