How startups should think about A.I. pitches

PreSeed Now brings you an in-depth profile of a different B2B or deep tech startup every Tuesday and Thursday. Subscribe to get it straight to your inbox.

Hello there,

If you’ve listened to our podcast, you’ll know that we’re very much up for trying things that expand our offering at PreSeed Now.

So today we’re switching things up and bit. Rather than talk to a UK-based startup, we’re talking to a pair of US-based investors who are immersed in the world in early-stage A.I. startups.

Brian and Chris from the Ride Home AI Fund have lots to say about how startups should be thinking about the hype cycle around artificial intelligence.

It’s a useful read for UK founders (or founders anywhere!) putting together a pitch to raise for their A.I. startup.

– Martin

Brian McCullough and Chris Messina on how founders can look beyond ‘A.I. for x’

If you follow the US tech scene closely, you might already be familiar with Brian McCullough and Chris Messina.

McCullough is the voice of the Techmeme Ride Home podcast, an internet historian, and sometimes guests on one of the world’s longest running tech podcasts, Leo Laporte’s This Week in Tech.

Messina is best known as the inventor of the hashtag, and has worked for the likes of Google and Uber. Oh, and he’s a prolific ‘hunter’ on Product Hunt, too. 

They’re also both investors in early-stage startups; McCullough through his rolling Ride Home Fund, and Messina as an angel. Now they’ve teamed up to launch the Ride Home AI Fund.

The fund is currently still raising, but it’s already actively investing in startups and counts Andreessen Horowitz’s Marc Andreessen and Chris Dixon among its existing limited partners. 

In an age where a bland press release can be spun up via ChatGPT in seconds, McCullough and Messina see their ‘value add’ as their communications and marketing experience, helping startups tell a compelling and unique story in the noisy field of A.I.

And the pair are particularly keen on finding experienced founders.

“Our bias is towards people for whom this isn't their first rodeo, and who say ‘this new technology allows me to do that one thing that I always wanted to do when I did my first startup, but wasn't possible before.’

Given the fund won’t limit itself to the US (“why can’t we invest in UK startups?” McCullough says), I thought it was worth getting their take on how A.I.-focused founders should think about product, investment, and more.

Going beyond ‘A.I. for x’

As with anyone who talks to a lot of early-stage startups in 2023, here at PreSeed Now we see a lot of ‘it’s like [existing tech] but with A.I.’ pitches.

This happens whenever an exciting new technology or business model emerges: ‘blockchain for x’, ‘Uber for x’… bolting a new thing onto an old thing seems like an easy win. But what founders quickly realise is that’s often not enough. 

It will generally be easier for incumbents to add generative A.I. to their own products than it will be for you to launch a whole new rival business that does the same thing.

“It's not enough to say that you're going to throw in an LLM or chat interface into software,” says Messina. 

“By this time next year, we will be shocked at the software that doesn't have some kind of generative A.I. aspect to it. And so, if you take that for granted, then the question is, how do you imagine this spooling out over the next five to seven years, which of course is like the time horizon for a fund like ours. 

“And so we have to be finding those founders that are thinking from a first principles perspective on how things are actually going to change and shift, as opposed to ‘oh, this is shiny and new’.”

McCullough adds that it’s about “thinking deeper about it and creating the things that weren't possible before, like Uber wasn't before the iPhone.”

Messina gives the example of Zoom’s new A.I. features, which he says instantly ‘Sherlocked’ a bunch of startups working on A.I.-based meeting summary tools. And McCullough points to Microsoft adding generative A.I. to apps like Excel. He says founders should be taking that reality to the next level: 

“Why even have a spreadsheet? What is possible now, that wasn't possible before? It's just software. What software is great at is enabling new things to be possible by abstracting away all of the complications.”

McCullough gives an example of FIY.ai, a startup he says the Ride Home AI Fund has invested in, which is training a large language model on user manuals for everything from printers, to cars, to refrigerators.

He explains that this startup could position itself as a tool for repair engineers, or as white label tech for OEMs’ own support offerings, or as a bot to help end users fix anything.

“That's something that wasn't possible before. This technology makes it possible. And it's training on the expertise of a niche and doing it better than the biggest model in the world.”

As another example, Messina suggests the compelling idea of ‘self-healing’ software that might soon be able to analyse its own code for bugs and write its own patches.

The mistakes A.I. startups make

What are some of the mistakes Messina and McCullough see A.I.-focused founders making?

First, there’s the old classic of creating the product before you know your market.

“We had an early conversation with these three Stanford kids that had discovered some of the generative A.I. architecture techniques, where you draw a napkin sketch, and it'll make an amazing-looking living room or something,” says Messina.

“It’s a great parlour trick and it's really cool. But they didn't know that Brian's wife is an architect. 

“So they’re proposing that this will be the Figma for architects. And the first question that we asked is ‘have you talked to any architects? What are the workflows that they actually have and what are the data formats and structures that they need to consider? And what are the compliance issues and how do you actually bring this to market?

“So even if you solve for hallucinations and you get brilliant outputs, are the outputs that are coming out of your software actually useful to solve real people's problems? And that's where you have to have that domain contact for it to make sense.”

Then there’s the mistake of assuming your ‘wow’ demo will be enough to lure investors at the pre-seed stage. 

“The macro environment makes it less attractive to invest in these spaces [around A.I,],” says Messina.

“Because people have been burned from all the crypto stuff. And also because from an interest rate perspective, the returns are going to be very different than they were when we were in a zero interest rate marketplace. 

“So if you're a founder, you might have something shiny and new and think you can just roll it out there and hope that people are gonna invest in you, because that's how it worked four years ago, 

“But that's not going to be the case anymore. So you've got a much better story in the pre-seed context. You need to really pull people in and go on a journey, as opposed to something that's like a kind of slot machine opportunity.”

Valuations: a view from the US

What kinds of a valuation should early-stage A.I. startups look at when raising pre-seed or Seed? These are US-centric valuation figures, but they give you an idea of what the market there looks like.

Some of these numbers might seem huge compared to the valuations many UK-based founders can achieve on their doorstep, but A.I. is an internationally competitive space so it’s worth knowing what investors in the US see.

“If you're looking to raise a pre-seed, you’ve got to be looking at $10 million and under [around £8 million], and even at Seed you're looking at $10 million to 15 million [$8 million to $12 million] but you’ve got to have like some compelling stuff there: founders, traction, other investors, things like that,” says McCullough.

He recalls investing in startups at far more inflated valuations a couple of years ago.

“The universe of £30 million to $50 million Seed and pre-seeds doesn't exist now. The best Seeds right now, even in this space, are topping out at 20 or 25 million. So that's a huge difference from two years ago, before the A.I. stuff.”

Keeping in touch with ‘Cerebral Valley’

If you’re thousands of miles from San Francisco’s Hayes Valley neighbourhood (nicknamed ‘Cerebral Valley’ because of all the A.I. activity there), McCullough believes you shouldn’t see that as a disadvantage.

“One of the things we've learned from history is that when there's a goldrush, people all go to the same place and there's a lot of groupthink that happens. And as we've seen in the last 20 years, that can have negative externalities and repercussions. 

“If I was a founder in Dublin, in Munich, in Nairobi, you could think of that as an advantage… In 2023 with everything distributed, being in a scene that is not San Francisco might give you a competitive advantage because you won't have the groupthink. 

“Whether that makes it harder for you to raise capital for your idea, that's one thing. But I think that, potentially, this is the opportunity that the tech industry hasn't had for 30 years, which is you can have different ideas and different scenes flower in a way that you couldn't before. 

“It will allow different scenes to create different paradigms and different ideas.”

But still, Messina says it’s worth founders making a visit to San Francisco if they can.

“It's super energising, it's very exciting. But there's, like, 40 or 50 events every month. It's really scattering and distracting. So when people go back to Lisbon, they go back to Berlin or wherever else they might be, they can actually focus on the work that they're doing. 

“It's sort of like a ‘work staycation’ or like going to Mecca, and then going back to wherever you actually live to do the work.”

Back on Tuesday

What did you think of today’s edition? Drop me a line and let me know.

On Tuesday, we’ll be back to our regular programming, profiling super-early-stage B2B and deep tech startups.

Join us in your inbox then to find out what we’ve got for you.