Will AI Replace Founders?

8 AI-Resistent Areas for Startups

It may be true that we’ve finally moved on from “software is eating the world” to “AI is eating software.”

And I mentioned to a friend this week that I think OpenAI should build a phone to compete with Apple since the need for apps may soon go away.

What does that mean for startups? Where are the actual opportunities?

IRL Products + Experiences

Ok, so we know it just became 1000x easier to write software.

But despite startups like Figure that have the potential to bring AI into the real world via advanced robotics, AI is currently still limited to the digital world.

While you may be half-jokingly worried that this means humans need to all become manual laborers, I disagree. Instead, we should leverage AI software tools to help accelerate our progress towards real-world breakthroughs such as:

• Nuclear fusion

• Quantum computing

• Energy storage

• Space exploration

• Gene editing

• IRL experiences for humans

• And, yes, augmented reality

I expect we’ll see some VCs publicly shift their strategy away from software as soon as the next few months.

Data

An AI is only as good as data it’s trained on. Not all data is publicly available on the internet, so even if an AI is able to search on its own it won’t have access to everything.

You could easily argue that startups are already valued based on the value of their proprietary data to a degree:

• Facebook → User interest and social graph

• Amazon → Purchase behavior

• Tesla → Sensor data around driving

We’re going to see the value of proprietary datasets dramatically increase. Startups that are able to train models on their own data, without giving up access to it, will be able to approach a market uniquely and potentially have an unbeatable advantage.

Unicorns will emerge around how to acquire, organize, and protect data sets. If I were building a B2B startup, this is where I’d look. Additionally, many more unicorns will emerge who figure out who do those three things for their specific company in various verticals.

Distribution, Brand, & Reputation

This Justin Kan quote has basically turned into an accepted, meme-able truth in Silicon Valley in the last 5 years.

It makes sense. A great product with no users loses to a worse product with scalable distribution every single time. Combined with the rise of no-code tools, creator platforms, ad saturation, and now AI we’ve seen products become easier to create while distribution becomes more competitive.

Rex Woodbury, a partner at Index Ventures who has built a combined 350k person following across Instagram, Twitter, LinkedIn, and a newsletter agrees and says it might not be a good thing.

Regardless, the best bet is to build this distribution channel through the founding team’s personal brands:

1. People connect with people more easily than brand accounts

2. Having a personal brand allows the distribution you build to travel with you as you build new companies throughout your career

Emotional Connections

Nike. Airbnb. McDonald’s.

The best brands make you feel something as soon as you hear their name or see their logo.

AI is going to take that to the next level.

The most successful and retentive AI-powered products will be the ones where users feel like they are actually forming an authentic emotional connection with the AI.

We’re already seeing this.

Replika’s decision to tone down the sexual messages from its “AI companion” app after pressure from some governments left some power users publicly contemplating on Reddit about how they could go on with their lives. Dark. Don’t Google it.

I have to imagine this is only the beginning of what we’ll see with this type of moat.

Exclusive Partnerships

As long as humans are pressing the buttons that makes AI do things, the decisions about who to work with (and which startups to kingmake) will still come down to “who do you want to work with?”

If AI allows anyone to create identical software quickly then a founder’s relationships will increasingly be the difference between gaining that distribution you need vs not.

Parts of a founder’s job like networking, sales, and partnerships will be much more resilient to AI than it may seem.

Unfortunately this may make it more difficult for founders who, like me, started off with few industry connections to break out unless they have truly novel ideas or other advantages.

Network Effects & Community

AI will make a network effects and community-driven moat less common but not less powerful. Basically, it will be harder to create network effects in the first place.

Imagine you find an attractive market to build a startup in. You have AI create some software and start marketing your product. Well, someone else sees your marketing and creates nearly identical software. They also start marketing. All of a sudden, you’ve got two fully formed competitors fighting it out very early on.

It’ll be challenging to reach the level of scale necessary for network effects to kick in. I worry you’ll simply need to outspend competitors, though I’m sure other paths to dominance will emerge.

My advice here for first movers? Do your best to avoid publicity if you find something truly interesting except with hyper-targeted marketing.

Patents & Regulations

Speed is a startup’s biggest advantage, and applying for patents and navigating complex regulatory environments can be big roadblocks.

But it seems likely that AI will be trained with a set of ethics and strict rules. This means it may not be easily able allow itself to take shortcuts around “the rules” like so many startups do in their early days.

Practically, this means that founders who are able to secure patents or work successfully in highly regulated markets will be better defended against AI-powered startups entering and gaining control of their market.

Help AI

In all of the situations mentioned above, there are viable startup ideas around helping AI conquer the barriers that currently exist.

To name one, helping AI interact with the real world like Figure is doing could result in many massive companies.

…or you could say forget the startup and build an AI church. Turns out they can be pretty profitable.

📚️