Did you first discover this newsletter via ChatGPT or Perplexity? An increasing number of PreSeed Now subscribers come from AI-based alternatives to traditional search engines.

Helping online publishers adjust to this shift is part of today’s startup’s offering. Read on to discover all about SUBJCT.

But first:

• North of England folks: YFM Equity Partners is running a ‘reverse pitch’ event next week that will let founders grill investors. I’ve always found events in this format to be valuable, so this one is worth a look. It takes place at the ABC Buildings in Manchester on Tuesday 29 April. [More info]

– Martin

Changing the SUBJCT: This startup wants to help publishers adjust to the AI search era

If you’ve ever been involved in digital publishing, you’ll know how important SEO is. Search engine optimisation is a key part of growing an audience for many publishers.

Amid a shift towards generative AI-based search results from Google, challengers like Perplexity, and people who use ChatGPT as a search engine, the goalposts are moving, but having your content fully optimised for search engines remains vitally important.

The problem is that in practice, search-optimising a massive library of online content is a huge pain. That’s because one of the main tasks to ensure you maximise page views is making sure all of your content contains links to other relevant pages on your website.

And if you have an enormous archive going back years, that’s an intimidating challenge.

This is where SUBJCT wants to come in, with a tool to make SEO teams, audience growth teams, and even journalists’ lives easier with a promise to, as its homepage puts it, “eliminate manual linking [and] unlock hundreds of hours of time savings and drive a 20% uplift in page impressions.”

Beyond SEO, a well-linked website can keep visitors engaged for longer, increasing page impressions in the process.

How it works

While SUBJCT plans more products in the future, its manual linking tool is its first offering. It analyses web pages to automatically figure out what other content to add links to.

The software identifies topics in pages and can, for example, link a mention of Donald Trump to a topic page containing all articles about him. It can also link mentions of particular events or topics to specific articles in the archive that can give readers more context.

“This is a very manual process at the moment for journalists, and it's a job they just do not want to do. They just don't do it - it's a totally broken process because they want to be writing great content. They don't want to be doing the linking,” says Ben Martin, co-founder of SUBJCT.

“And SEO teams don't have the ability to scale [linking] across big content archives because they're generally a smaller team within the organisation.”

Martin says the sophistication of SUBJCT comes from its ability to do this linking task at great scale.

“We have the ability to analyse and do entity analysis, and named entity recognition on archives which go from 1,000 articles up to millions of articles. When you've got potentially hundreds of thousands, or a million articles within a website archive, it becomes very hard to scale this.”

Meanwhile, Martin says SUBJCT uses vector embeddings to underpin its semantic capabilities. This is technology helps establish the meaning of words beyond a simple keyword match, making it possible to add links to highly relevant content, rather than just something with the same words in it.

“The keyword approach to SEO is becoming obsolete. It's more about topic clusters, semantic understanding. That’s what we're trying to deliver with this product,” says Martin.

The software is API-first, designed to integrate with large publishers’ custom content management systems. Given that WordPress remains a significant player in the professional publishing space, SUBJCT is currently working on a plugin for the platform.

The story so far

Martin and his fellow co-founder at SUBJCT, Dan Reeves, have known each other for a long time while working in similar roles in the UK media industry. Most recently, they ran a digital content and marketing agency together.

Martin explains that while working for clients, the problem of making the most of large content archives kept coming up:

“Brands were producing a huge amount of content, but this whole process of optimising your content, the internal linking, kept coming back as a broken process. There's a lot of unearth value in content archives, and this is what we're trying to address.”

Reeves says they began to address the problem with the help of funding from the Google Digital News Initiative:

“We had this pain point keep coming back to us, so we decided to launch this business purely around an API-first, B2B SaaS product to essentially eliminate a pain point, but also drive content engagement within this era of AI search.

“This space has always been exciting to us. We know the marketplace, we've got extensive relationships with news publishers and brand publishers. So it's just part of our DNA.”

Martin says SUBJCT is currently in testing with 10 publishers in the UK, USA, and Australia to validate and further develop the product. In the future, the startup is looking at how it can expand beyond text, to video and audio content. They’re also looking into creating knowledge graphs from its customers’ content.

“That’s way you get a source of truth for your brand in the world of large language models,” Martin says.

Subscribe to Premium to read the rest.