‘Vibe coding’ must be up there as a contender for new tech term of the year.

But today’s startup goes beyond what trendy tools like Replit can do, to help product managers ship AI-focused, production-ready features more quickly with LLM prompts, without having to wait for developer capacity.

Scroll down to read all about Rightbrain.

But first:

• At PreSeed Now we’re currently filling up our schedule through late May and June, to early July. We don’t want to miss anything!

• Know a UK-based B2B or deep tech startup anywhere from inception to raising seed that we should be talking to and potentially featuring? Our door is open!

• Email me: [email protected] with your hot tips or to pitch your own startup.

– Martin

Rightbrain wants to make it easier to build AI features into existing apps… with the help of AI

The term ‘vibe coding’ was coined just three months ago by Andrej Karpathy as “a new kind of coding… where you fully give in to the vibes, embrace exponentials, and forget that the code even exists.”

The term was initially intended to mean a kind of casual, weekend project type of development using AI. But it’s quickly been adapted to refer to more serious coping efforts too.

Y Combinator CEO Garry Tan has said vibe coding will build the next generation of lean, successful startups. But while tools ChatGPT like can come up with code, and tools like Replit can build a simple app or website for you, one UK startup believes there’s room to go further.

Rightbrain pitches itself as being focused on helping users “Ship AI features faster. From prompt to production.”

What does this look like in practice? Co-founder Peter Cheyne explains:

“We allow people to build production-ready generative AI features in natural language. We take care of the actual generation of the APIs themselves, we do the hosting, we do the security.

“We’re about getting LLMs to perform in a manner that is consistent, easy to integrate, easy to host, and implementing that in such a way that a team who's never actually worked with this before feels comfortable that they are using it in the right way, they are hosting it in the right way, the data's been handled the right way.

“That's daunting for a lot of businesses. And the problem tends to be that often the people who understand the business problem aren't the ones who deliver the final solution. So you end up with a disconnect between the potential application of generative AI and the actual real world deployment of it. And that's where we come in.”

Rather than explain in detail how it works, I’ll defer to this video the Rightbrain team put together, which outlines the process:

Cheyne explains that Rightbrain is designed so less technical product managers can articulate in natural language an AI-based feature that they want to build. The system takes that, packages it up, sends it to a relevant LLM, and turns it into code that will work reliably in the app that production manager is responsible for.

“It allows you, in a pretty quick way, to get prototyping and to test whether this is going to be useful for your use case, and if it is, you can just generate an API endpoint that you can then pass on to your technical team to integrate,” Cheyne says.

So essentially, a product manager can build a feature that relies on LLMs and then simply pass it on to the developers as an API call they can build into the app. The API call goes to Rightbrain, which then handles the data going to and from the LLM “in a recorded, safe, and clean manner,” Cheyne’s fellow Rightbrain co-founder Matt Wells adds.

The story so far

Cheyne and Wells met while working at one of the big names of Newcastle tech, Performance Horizon Group, now known as Partnerize, where Cheyne was a co-founder.

The pair then worked together again after Cheyne founded Bottlepay, a consumer-focused cryptocurrency app that was acquired by NYDIG, although Wells also spent time working for GitHub.

And then they decided to work together a third time, this time as co-founders of Rightbrain, which they started to build in late 2023.

Rightbrain launched to the public last month and Cheyne says it has clocked up its first 100 user accounts.

“We're really big proponents of product-led growth, so we're in conversations with a lot of our early adopters to better understand how they are using the product, and that's dictating what we're doing next,” Cheyne says.

One direction they’re exploring is integration with third-party tools in the workflow automation space, to make discovering and using Rightbrain easier without having to switch away from an app a user is already working with. They’re also looking at how they can integrate agentic AI into their product offering.

Subscribe to Premium to read the rest.