In partnership with

AI coding tools are transforming the pre-seed startup landscape.

Just as the introduction of drum machines meant bands didn’t have to search for a drummer before they could properly get going, ‘vibe coding’ gives founders a head start without having to hunt down a technical co-founder.

I recently visited a startup-building event themed around AI tools to find out more. Read on below…

But first:

• Congratulations to my former colleague, Laura IH Bennett, who has co-authored a book: Developing Entrepreneurial Ecosystems: Place, Processes, People

• The book is based on Laura’s PhD work, and it looks like it will be really useful to many people who read PreSeed Now.

• Laura has more info here.

   

– Martin

Payroll errors cost more than you think

While many businesses are solving problems at lightspeed, their payroll systems seem to stay stuck in the past. Deel's free Payroll Toolkit shows you what's actually changing in payroll this year, which problems hit first, and how to fix them before they cost you. Because new compliance rules, AI automation, and multi-country remote teams are all colliding at once.

Check out the free Deel Payroll Toolkit today and get a step-by-step roadmap to modernize operations, reduce manual work, and build a payroll strategy that scales with confidence.

Download the Toolkit for free today

AI coding is transforming startups. But how far should a founder go with it?

AI coding tools have transformed software developers’ jobs shockingly quickly. They’ve also opened up opportunities for founders kicking off a new startup to get further without any technical skills.

Gone are the days of the founder without a CTO having to hire a tech agency to build their prototype. Learn to use a tool like Lovable or Bolt and you can quickly have a workable tool to get your idea off the ground and start attracting users, customers, and investors.

But how much can you really do with these AI tools? And how much should you do with them if you’re serious about building a scalable, VC-backed business?

To get some answers, I spoke to founders and investors at Build Like A Pro, an event that took place in Manchester last month, organised by Antler and Glenluna Ventures.

With tracks for first-time builders and experienced builders, the attendees were a mix of established early-stage startups and founders right at the start of their journeys.

Many of them were attracted by the access to investors and the chance to work alongside other founders in a structured environment. A good number of them said the Lovable credits that were on offer to attendees were a draw too.

Changing founders’ lives

Susie Kilcoyne is the founder of Locket, a startup billed as “AI-powered commerce infrastructure that creates an individual storefront for each customer in real time”.

Kilcoyne tells me she created 1.4 million lines of code with Lovable in the past six months and is completely sold on the benefits of AI coding tools to early-stage founders.

“As cringey as it sounds, it has changed my life… It has done things that a developer told me would never be able to happen."

She gives the example of a marketing email that can be updated after it has been sent so that it is as well-targeted and up-to-date as possible when it is opened.

“I asked a developer, can I create live email? Can we update that email in real time before they've opened it? They said no, Lovable said yes, and we're now testing it, and it's been possible to do that. Lovable should almost be my co-founder.

“Every day I'm shipping multiple things, especially for our beta customers.”

There was a similar sentiment from Zsike Peter, who was at Build Like A Pro to work on CoMamma, a platform to help single and solo-by-choice mothers find others in the same situation to support each other.

She finds Lovable useful not just for creating what she wants, but for suggesting ideas to make her product better.

“It's super easy, to the point where it explains why something isn't working. It makes improvement suggestions that I haven't thought of through prompting independently and researching independently through ChatGPT, or even by asking users. It comes up with suggestions where I think ‘okay, yeah, I should have thought of that’.”

Peter must be doing something right with this approach, as she won the event’s ‘Build’ track for people just getting started with AI coding tools.

Then I bumped into a familiar face, Dan Bottomley from voxANN, the media localisation startup we profiled here at PreSeed Now a year ago.

voxANN was at the event to give the team a push to develop some of the ‘fun’ things on their roadmap that had got stuck in the development queue behind more critical features like security and billing.

Bottomley said that in their two days at the event, they had developed a tool to automatically analyse screenplays and extract information that is important to voxANN users. It works with their existing tech to align with audio in corresponding media files.

As a co-founder without a coding background, Bottomley found working with AI coding tools to be an enlightening experience.

“I’ve learned I can do quite a lot with vibe coding. It surprised me what it can do in the backend as well as the frontend. And the more I actually watch what it's doing, and thinking, and working on, the more I feel like I'm actually learning about how the tech works.

“I’m from an audio background and I understand the mechanics of that, but I feel like I'm now starting to understand the mechanics of the software more.”

How far should you scale with AI code?

Getting something off the ground with AI-generated code is one thing. But is it best thought of as a prototyping tool to prove product-market fit before you build the ‘real’ software the traditional way?

Or can you go further?

Antler was a co-organiser of Build Like A Pro, and is also an investor in Lovable, so certainly has a vested interest in AI coding tools being widely used. But he is balanced in his views about how far they should take a founder… for now.

David Houghton, Principal at Antler, was at the event and I asked him for his perspective on the tools. (We previously interviewed David as part of his other role running Social Mobility Ventures).

He says the emergence of AI coding is a mixed blessing when it comes to an investor’s inbox.

“From an investor point of view, we're just seeing so many more startups emerging these days because the barrier to entry is so low… We're seeing lots of companies who are building great, valuable businesses at the early stage that just couldn't exist five years ago.

“The flip side of that is it’s very difficult to determine who are going to be the winners, because the barrier to entry being so low means that the competition is much higher. So what's the edge that we need to look for to find the winners? It’s much more difficult.”

Building Locket the right way to scale it is something very much on Kilcoyne’s mind, even though she’s currently enjoying the speed of creating code with AI.

“My technical advisor said that for now I can keep using Lovable so I can be fast, because I ship constantly. He said I should build the frontend in Lovable and start building the backend on Azure or AWS, so that I can still move fast but I’m building the foundations for when I do need to move over [to a more scalable infrastructure].

“The main question our initial customers had was about security. But from his perspective, and from my perspective, the research I've done on Lovable is it's perfect at this early stage. So he said I'll probably get a year out of it doing this. I'll wait and see how it goes.”

Houghton agrees that knowing when to switch over to a more traditional tech architecture is critical.

“If you are building solely with AI code generators, that's probably a bit of a red flag at some point. There comes a moment where you need to have technical expertise in house to really scale something.

“I'm not sure that we're quite there yet with Lovable or Cursor where you can build a product that services tens of thousands of users and enterprise customers yet, but I think we'll get there.”

David Levine of Glenluna Ventures, who organised Build Like A Pro, says plans are already afoot for the next event.

What are your thoughts on building a startup with AI code? Whether you’re an investor or a founder, drop me a line and let me know how it’s helping, or hindering, your work. It would be fun to explore this topic further in another piece soon.