A new ecosystem of AI-powered products

What is surprising quickly becomes commonplace. It's been less than two years since OpenAI released his ChatGPT to the unknown world, sparking a global discussion about the possibilities, opportunities, and dangers associated with generative artificial intelligence. Last month, the company announced its latest upgrades that promise significant performance improvements, and while it received considerable media coverage, it was mostly about the technology itself, not its impact. Although still in its infancy, generative AI seems strangely mainstream.

And tech entrepreneurs are increasingly looking to embed the technology in their own products. As a journalist writing about start-ups, I get around 30-40 emails a day from young tech companies looking to grab media attention. They are increasingly offering “AI-driven” products. And that begs the question: if AI relies on breakthrough technology backed by the financial muscle of giants like the Googles and Microsofts of the world, how can less-funded companies get in on the action? Last week, I spoke to two UK company founders who have put artificial intelligence at the core of their products to find out.

In January this year, London-based legal services business RobinAI successfully raised $26 million in a Series B round led by Temasek. The company, founded in 2019 by lawyer Richard Robinson, describes its product as a “legal co-pilot” focused on business contracts. More precisely, it is a platform that uses generative AI to enable enterprise customers to quickly review contracts and create new ones.

CEO Richard Robinson, an alumnus of the law firm Clifford Chance, started work on a chatbot-to-legal platform aimed at helping tenants get their deposits back from landlords at the end of a property lease, which can be difficult. “Tenants can talk to a bot about their situation, and the bot will generate some scary legal documentation for the landlord,” he says.

In contrast, RobinAI, co-founded by Robinson and machine learning researcher James Clough, focuses on the B2B market. In addition to helping small and medium-sized businesses review and create contracts, it also serves large enterprises, including those on the Fortune 500 list. The company claims it can reduce the time it takes to review contracts by around 80%.

Do more with less

As Robinson explains, Robin AI's services are primarily pitched to in-house legal teams. “They're often asked to do more with less,” he says. But the company also pitches to CEOs, CIOs and CFOs. He acknowledges that it's not always an easy sell. “There was a lot of doubt,” he says. “A lot of people were very skeptical about AI.”

To counter that skepticism, RobinAI has in-house lawyers who work with clients to use the software, which has helped build trust in the platform's ability to deliver robust legal services.

But how much of this is actually due to AI? For years, and long before generative AI, law firms have been automating some, albeit relatively simple, functions. We've been talking about it. But Robinson says AI has changed the game. “The big area that has changed is the ability of large language models to generate new text,” he says. In addition to text generation, the model also has the ability to extract information from long documents.

The question is, how can small companies offering proprietary technology afford to swim in the big pond of language models? For Robin AI, the answer lies in collaborating with others. The company developed its own legal AI product while partnering with AI “research and safety” company Anthropic on language models.

Skill Combinations

For many companies, the key to creating AI-based services is to combine specialized in-house skills, such as law, education, and cyber security, with third-party models. And there are options.

For example, cybersecurity startup Appella chose an open source model to underpin its technology. The company's flagship product is an “AI receptionist,” as CEO and co-founder James Drayson describes it. From a cybersecurity perspective, its role is to scan incoming calls for evidence of potential fraud. However, you can also use the same technology to streamline your workflow and manage calls more efficiently.

Appella is in an earlier stage of entrepreneurship than RobinAI. To date, the company has raised a pre-seed round of funding and has also won an Innovation His Award grant. Having completed his beta testing and Appella currently in talks with potential customers, Drayson sees great potential in the product. “There are so many industries that we serve,” he says Drayson. “Any business can fall victim to fraud.”

To develop this product, Appella eschewed the approach of connecting to its own generative AI platform through an API. Drayson said open source solutions have provided the best way to create products that can be honed to meet customer needs. Data training was performed on open-source LLM using a database containing information about scams and scams.

An ecosystem around AI is developing, with startups leveraging commercially available software and open source alternatives to power their own solutions. This opens the door to more AI-powered technology products.